Root cause analysis method and device, electronic equipment and storage medium

Abstract

The invention relates to a root cause analysis method and device, electronic equipment and a storage medium, and belongs to the technical field of root cause analysis. The method comprises the steps that a target index of a product and candidate dimension items of the target index under multiple candidate dimensions are acquired, and all the candidate dimension items of the target index under any candidate dimension are used for being aggregated into the target index; based on the data of the plurality of candidate dimension items, screening out a first dimension item from the plurality of candidate dimension items to serve as a single-dimensional root cause analysis result; based on the data of the plurality of candidate dimension items, screening out a combination of second dimension items from combinations of the plurality of candidate dimension items to serve as a multi-dimensional root cause analysis result; and obtaining a root cause analysis result of the target index based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result. Therefore, the one-dimensional root cause analysis result and the multi-dimensional root cause analysis result are comprehensively considered, the root cause analysis result of the target index is obtained, and the comprehensiveness of root cause analysis is improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of root cause analysis, and in particular to a root cause analysis method, device, electronic device and storage medium. Background Art

[0002] At present, root cause analysis technology has been widely used in the fields of product quality monitoring, product indicator monitoring, data monitoring, etc. For example, in product quality monitoring, root cause analysis can be performed on product quality to find the specific process and specific raw materials that most affect product quality, and optimize and improve the specific process and specific raw materials to improve product quality. However, the root cause analysis method in the relevant technology has the problem that the root cause analysis results are relatively single. Summary of the invention

[0003] The present disclosure provides a root cause analysis method, device, electronic device, computer-readable storage medium, and computer program product to at least solve the problem that the root cause analysis method in the related art has a relatively single root cause analysis result. The technical solution of the present disclosure is as follows:

[0004] According to a first aspect of an embodiment of the present disclosure, a root cause analysis method is provided, including: obtaining a target indicator of a product, and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator; based on data of the multiple candidate dimension items, screening out a first dimension item from the multiple candidate dimension items as a single-dimensional root cause analysis result of the target indicator; based on data of the multiple candidate dimension items, screening out a combination of second dimension items from a combination of the multiple candidate dimension items as a multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension; based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator, obtaining a root cause analysis result of the target indicator.

[0005] In one embodiment of the present disclosure, the method of filtering out the first dimension item from the multiple candidate dimension items based on the data of the multiple candidate dimension items includes: determining a sixth dimension to be deleted from the multiple candidate dimensions based on the number of candidate dimension items under the candidate dimensions, the name of the candidate dimensions, and at least one of the user's preference for the candidate dimensions; deleting the sixth dimension from the multiple candidate dimensions, and determining the multiple candidate dimensions remaining after the deletion as the first dimension; and filtering out the first dimension item from the multiple candidate dimension items under the first dimension based on the data of the multiple candidate dimension items under the first dimension.

[0006] In one embodiment of the present disclosure, the sixth dimension to be deleted is determined from a plurality of candidate dimensions based on at least one of the number of candidate dimension items under the candidate dimension, the name of the candidate dimension, and the user's preference for the candidate dimension, including: if the number of candidate dimension items under the candidate dimension is greater than or equal to a first set threshold, or the number of candidate dimension items under the candidate dimension is less than or equal to a second set threshold, the candidate dimension is determined as the sixth dimension; and / or, if the name of the candidate dimension is a set name, the candidate dimension is determined as the sixth dimension; and / or, if the user's preference for the candidate dimension is less than or equal to a third set threshold, the candidate dimension is determined as the sixth dimension.

[0007] In one embodiment of the present disclosure, the method of selecting a first dimension item from the plurality of candidate dimension items based on the data of the plurality of candidate dimension items includes: obtaining the explanatory power and surprise degree of the candidate dimension item based on the data of the candidate dimension item; sorting the plurality of candidate dimension items from large to small according to the absolute value of the explanatory power, adding the top N candidate dimension items to the target set, and / or adding the candidate dimension items whose absolute value of the explanatory power is greater than a fourth set threshold to the target set, wherein N is a positive integer; sorting the plurality of candidate dimension items in the target set from large to small according to the absolute value of the surprise degree, and determining the top M candidate dimension items as the first dimension items, wherein M is a positive integer.

[0008] In one embodiment of the present disclosure, the method of selecting a first dimension item from the plurality of candidate dimension items based on the data of the plurality of candidate dimension items includes: obtaining the contribution of the candidate dimension item based on the actual value and reference value of the candidate dimension item at the current moment, and the actual value and reference value of the target indicator at the current moment; sorting the plurality of candidate dimension items from large to small according to the absolute value of the contribution, and determining the first M candidate dimension items as the first dimension items, where M is a positive integer.

[0009] In one embodiment of the present disclosure, the method of screening out combinations of second dimension items from combinations of multiple candidate dimension items based on data of multiple candidate dimension items includes: screening out combinations of third dimension items of the i-th round from combinations of multiple candidate dimension items based on data of multiple candidate dimension items, wherein i is a positive integer and the initial value of i is 1; judging whether the number of combinations of third dimension items of the i-th round is less than or equal to a fifth set threshold; if the number of combinations of third dimension items of the i-th round is less than or equal to the fifth set threshold, taking the combinations of third dimension items of the i-th round as combinations of fourth dimension items, and obtaining combinations of second dimension items based on the combinations of fourth dimension items; if the number of combinations of third dimension items of the i-th round is greater than the fifth set threshold, continuing the screening process of the i+1th round until the number of combinations of third dimension items of the target round is less than or equal to the fifth set threshold, taking the combinations of third dimension items of the target round as combinations of fourth dimension items, and obtaining combinations of second dimension items based on the combinations of fourth dimension items.

[0010] In one embodiment of the present disclosure, based on the data of the plurality of candidate dimension items, a combination of third dimension items of the i-th round is screened out from the combination of the plurality of candidate dimension items, including: if i=1, based on the data of the first dimension items under any two candidate dimensions, a combination of third dimension items of the first round is screened out from the combination of the first dimension items under any two candidate dimensions.

[0011] In one embodiment of the present disclosure, based on the data of the multiple candidate dimension items, a combination of third dimension items of the i-th round is screened out from the combination of the multiple candidate dimension items, including: if i≥2, splitting the combination of third dimension items of the i-1-th round to obtain multiple third dimension items of the i-1-th round; using the multiple third dimension items of the i-1-th round as fifth dimension items; based on the data of the fifth dimension items under any i+1 candidate dimensions, a combination of third dimension items of the i-th round is screened out from the combination of the fifth dimension items under any i+1 candidate dimensions.

[0012] In one embodiment of the present disclosure, obtaining the combination of the second dimension items based on the combination of the fourth dimension items includes: splitting the combination of the fourth dimension items to obtain multiple fourth dimension items; and filtering out the combination of the second dimension items from the combination of the multiple fourth dimension items based on the data of the multiple fourth dimension items.

[0013] In one embodiment of the present disclosure, the data of the candidate dimension item includes a reference value of the candidate dimension item at the current moment, and the method further includes: obtaining actual values ​​of the candidate dimension item at multiple historical moments; taking the average of the actual values ​​of the candidate dimension item at multiple historical moments as the reference value of the candidate dimension item at the current moment; or, performing weighted summation of the actual values ​​of the candidate dimension item at multiple historical moments to obtain the reference value of the candidate dimension item at the current moment.

[0014] In one embodiment of the present disclosure, the root cause analysis results of the target indicator are obtained based on the unidimensional root cause analysis results and the multidimensional root cause analysis results of the target indicator, including: normalizing the contribution of the unidimensional root cause analysis results and the multidimensional root cause analysis results of the target indicator; sorting the unidimensional root cause analysis results and the multidimensional root cause analysis results of the target indicator from large to small according to the absolute values ​​of the contribution after normalization, and taking the first Q results of the sorting as the root cause analysis results of the target indicator, wherein Q is a positive integer.

[0015] In one embodiment of the present disclosure, the contribution of the multidimensional root cause analysis result of the target indicator is normalized, including: splitting the combination of the second dimension items to obtain multiple second dimension items; updating the contribution of the multidimensional root cause analysis result of the target indicator based on the number of candidate dimensions to which the multiple second dimension items belong; and normalizing the updated contribution of the multidimensional root cause analysis result of the target indicator.

[0016] According to a second aspect of an embodiment of the present disclosure, a root cause analysis device is provided, comprising: a first acquisition module, configured to execute acquisition of a target indicator of a product, and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator; a first screening module, configured to execute data based on the multiple candidate dimension items, and screen out first dimension items from the multiple candidate dimension items as a single-dimensional root cause analysis result of the target indicator; a second screening module, configured to execute data based on the multiple candidate dimension items, and screen out a combination of second dimension items from a combination of the multiple candidate dimension items as a multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension; and a second acquisition module, configured to execute single-dimensional root cause analysis results and multi-dimensional root cause analysis results based on the target indicator to obtain a root cause analysis result of the target indicator.

[0017] In one embodiment of the present disclosure, the first screening module is further configured to perform: determining a sixth dimension to be deleted from a plurality of the candidate dimensions based on at least one of the number of candidate dimension items under the candidate dimensions, the name of the candidate dimensions, and the user's preference for the candidate dimensions; deleting the sixth dimension from the plurality of the candidate dimensions, and determining the plurality of candidate dimensions remaining after the deletion as the first dimension; and screening out the first dimension items from the plurality of candidate dimension items under the first dimension based on the data of the plurality of candidate dimension items under the first dimension.

[0018] In one embodiment of the present disclosure, the first screening module is further configured to execute: if the number of candidate dimension items under the candidate dimension is greater than or equal to a first set threshold, or the number of candidate dimension items under the candidate dimension is less than or equal to a second set threshold, the candidate dimension is determined as the sixth dimension; and / or, if the name of the candidate dimension is a set name, the candidate dimension is determined as the sixth dimension; and / or, if the user's preference for the candidate dimension is less than or equal to a third set threshold, the candidate dimension is determined as the sixth dimension.

[0019] In one embodiment of the present disclosure, the first screening module is further configured to perform: based on the data of the candidate dimension items, obtaining the explanatory power and surprise degree of the candidate dimension items; sorting the multiple candidate dimension items from large to small according to the absolute value of the explanatory power, adding the top N candidate dimension items to the target set, and / or adding the candidate dimension items whose absolute value of the explanatory power is greater than a fourth set threshold to the target set, wherein N is a positive integer; sorting the multiple candidate dimension items in the target set from large to small according to the absolute value of the surprise degree, and determining the top M candidate dimension items in the target set as the first dimension items, wherein M is a positive integer.

[0020] In one embodiment of the present disclosure, the first screening module is further configured to perform: obtaining the contribution of the candidate dimension item based on the actual value and reference value of the candidate dimension item at the current moment, and the actual value and reference value of the target indicator at the current moment; sorting the multiple candidate dimension items from large to small according to the absolute value of the contribution, and determining the first M candidate dimension items in the sorting as the first dimension items, where M is a positive integer.

[0021] In one embodiment of the present disclosure, the second screening module is further configured to perform: based on the data of the plurality of candidate dimension items, screening out the combination of the third dimension items of the i-th round from the combination of the plurality of candidate dimension items, wherein i is a positive integer and the initial value of i is 1; judging whether the number of the combination of the third dimension items of the i-th round is less than or equal to the fifth set threshold; if the number of the combination of the third dimension items of the i-th round is less than or equal to the fifth set threshold, taking the combination of the third dimension items of the i-th round as the combination of the fourth dimension items, and obtaining the combination of the second dimension items based on the combination of the fourth dimension items; if the number of the combination of the third dimension items of the i-th round is greater than the fifth set threshold, continuing to perform the screening process of the i+1th round until the number of the combination of the third dimension items of the target round is less than or equal to the fifth set threshold, taking the combination of the third dimension items of the target round as the combination of the fourth dimension items, and obtaining the combination of the second dimension items based on the combination of the fourth dimension items.

[0022] In one embodiment of the present disclosure, the second screening module is further configured to execute: if i=1, based on the data of the first dimension items under any two candidate dimensions, a combination of the third dimension items of the first round is screened out from the combination of the first dimension items under any two candidate dimensions.

[0023] In one embodiment of the present disclosure, the second screening module is further configured to perform: if i≥2, split the combination of the third dimension items of the i-1th round to obtain multiple third dimension items of the i-1th round; use the multiple third dimension items of the i-1th round as fifth dimension items; based on the data of the fifth dimension items under any i+1 candidate dimensions, screen out the combination of the third dimension items of the i-1th round from the combination of the fifth dimension items under any i+1 candidate dimensions.

[0024] In one embodiment of the present disclosure, the second screening module is further configured to execute: splitting the combination of the fourth dimension items to obtain a plurality of fourth dimension items; and screening out the combination of the second dimension items from the combination of the plurality of fourth dimension items based on the data of the plurality of fourth dimension items.

[0025] In one embodiment of the present disclosure, the data of the candidate dimension item includes a reference value of the candidate dimension item at the current moment, and the first acquisition module is further configured to execute: obtaining actual values ​​of the candidate dimension item at multiple historical moments; taking the average of the actual values ​​of the candidate dimension item at multiple historical moments as the reference value of the candidate dimension item at the current moment; or, performing weighted summation of the actual values ​​of the candidate dimension item at multiple historical moments to obtain the reference value of the candidate dimension item at the current moment.

[0026] In one embodiment of the present disclosure, the second acquisition module is further configured to execute: normalizing the contribution of the single-dimensional root cause analysis results and the multi-dimensional root cause analysis results of the target indicator; sorting the single-dimensional root cause analysis results and the multi-dimensional root cause analysis results of the target indicator from large to small according to the absolute value of the contribution after normalization, and taking the first Q results of the sorting as the root cause analysis results of the target indicator, where Q is a positive integer.

[0027] In one embodiment of the present disclosure, the second acquisition module is further configured to execute: splitting the combination of the second dimensional items to obtain multiple second dimensional items; updating the contribution of the multidimensional root cause analysis result of the target indicator based on the number of candidate dimensions to which the multiple second dimensional items belong; and normalizing the updated contribution of the multidimensional root cause analysis result of the target indicator.

[0028] According to a third aspect of an embodiment of the present disclosure, there is provided an electronic device, comprising a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to implement the steps of the method described in the first aspect of the embodiment of the present disclosure.

[0029] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which computer program instructions are stored. When the program instructions are executed by a processor, the steps of the method described in the first aspect of the embodiment of the present disclosure are implemented.

[0030] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor of an electronic device, the computer program implements the steps of the method described in the first aspect of the embodiment of the present disclosure.

[0031] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects: a single-dimensional root cause analysis can be performed on the target indicator to obtain first-dimensional items, and a multi-dimensional root cause analysis can be performed on the target indicator to obtain a combination of second-dimensional items. The root cause analysis results of the target indicator are obtained by comprehensively considering the single-dimensional root cause analysis results and the multi-dimensional root cause analysis results, thereby improving the comprehensiveness of the root cause analysis.

[0032] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The drawings herein are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description are used to explain the principles of the present disclosure, and do not constitute improper limitations on the present disclosure.

[0034] Figure 1 The present invention is a flow chart showing a root cause analysis method according to an exemplary embodiment.

[0035] Figure 2 The present invention is a flow chart showing a root cause analysis method according to another exemplary embodiment.

[0036] Figure 3 The present invention is a flow chart showing a root cause analysis method according to another exemplary embodiment.

[0037] Figure 4 The present invention is a flow chart showing a root cause analysis method according to another exemplary embodiment.

[0038] Figure 5 The present invention is a flow chart showing a root cause analysis method according to another exemplary embodiment.

[0039] Figure 6 The invention is a block diagram of a root cause analysis device according to an exemplary embodiment.

[0040] Figure 7 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0041] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings.

[0042] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0043] The acquisition, storage, use, and processing of data in the technical solution disclosed in this disclosure are in compliance with the relevant provisions of national laws and regulations.

[0044] Figure 1 is a flow chart of a root cause analysis method according to an exemplary embodiment. Figure 1 As shown, the root cause analysis method of the embodiment of the present disclosure includes the following steps.

[0045] S101, obtaining a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator.

[0046] It should be noted that the execution subject of the root cause analysis method of the embodiment of the present disclosure is an electronic device, and the electronic device includes a mobile phone, a notebook, a desktop computer, a vehicle-mounted terminal, a smart home appliance, a wearable device, etc. Among them, the wearable device may include a wrist-worn device (such as a smart watch, a smart bracelet), a head-worn device, a foot-worn device, etc. The root cause analysis method of the embodiment of the present disclosure can be executed by the root cause analysis device of the embodiment of the present disclosure, and the root cause analysis device of the embodiment of the present disclosure can be configured in any electronic device to execute the root cause analysis method of the embodiment of the present disclosure.

[0047] It should be noted that the target indicator refers to the indicator of the product to be root cause analyzed, the candidate dimension refers to the dimension for root cause analysis of the target indicator, and the candidate dimension item refers to the result obtained by splitting the target indicator according to the candidate dimension. A candidate dimension can correspond to at least one candidate dimension item, and different candidate dimensions can correspond to different candidate dimension items. All candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator, that is, the target indicator is obtained by aggregating all candidate dimension items of the target indicator under any candidate dimension.

[0048] It should be noted that the data aggregation process can be implemented by adopting at least one data aggregation method in the relevant technology, and the sum, maximum value, minimum value, average value, variance, etc. of the actual values ​​of all candidate dimension items of the target indicator under any candidate dimension can be used as the actual value of the target indicator.

[0049] It should be noted that there are no excessive restrictions on products, target indicators, candidate dimensions, and candidate dimension items. For example, products may include mobile phones, food, financial products, etc., target indicators may include profit, cost, sales volume, sales amount, product quality, product operating parameters, etc., and candidate dimensions may include administrative divisions (such as provinces), sales channels (such as e-commerce platforms), sales stores, processes, raw materials, etc.

[0050] For example, taking the product as mobile phone and the target indicator as national sales amount, the candidate dimension items of the national sales amount under administrative divisions include the sales amount of mobile phones in Province A, the sales amount of mobile phones in Province B, the sales amount of mobile phones in Province C, and the sales amount of mobile phones in Province D; the candidate dimension items of the national sales amount under sales channels include the sales amount of mobile phones in sales channel 1, the sales amount of mobile phones in sales channel 2, and the sales amount of mobile phones in sales channel 3.

[0051] The sales amount of mobile phones in province A, the sales amount of mobile phones in province B, the sales amount of mobile phones in province C, and the sales amount of mobile phones in province D are aggregated into the national sales amount of mobile phones. For example, the actual value of the national sales amount of mobile phones can be determined as the sum of the actual values ​​of the sales amount of mobile phones in province A, the sales amount of mobile phones in province B, the sales amount of mobile phones in province C, and the sales amount of mobile phones in province D.

[0052] The sales amount of mobile phones in sales channel 1, the sales amount of mobile phones in sales channel 2, and the sales amount of mobile phones in sales channel 3 are aggregated into the national sales amount of mobile phones. For example, the actual value of the national sales amount of mobile phones can be determined as the sum of the actual values ​​of the sales amount of mobile phones in sales channel 1, the sales amount of mobile phones in sales channel 2, and the sales amount of mobile phones in sales channel 3.

[0053] S102, based on the data of the plurality of candidate dimension items, a first dimension item is screened out from the plurality of candidate dimension items as a single-dimensional root cause analysis result of the target indicator.

[0054] It should be noted that there are no excessive restrictions on the data of candidate dimension items. For example, it may include the actual value and reference value of the candidate dimension item at the current moment, the actual value and reference value of the candidate dimension item at multiple historical moments, etc. Among them, the historical moment is earlier than the current moment. For example, if the current moment is August 10, the historical moments may include August 9, August 8, August 7, August 6, etc.

[0055] In one embodiment, the data of the candidate dimension items include reference values ​​of the candidate dimension items at the current moment, and the method further includes obtaining actual values ​​of the candidate dimension items at multiple historical moments, taking the average of the actual values ​​of the candidate dimension items at multiple historical moments as the reference value of the candidate dimension items at the current moment, or performing weighted summation on the actual values ​​of the candidate dimension items at multiple historical moments to obtain the reference value of the candidate dimension items at the current moment. Therefore, in this method, the actual values ​​of the candidate dimension items at multiple historical moments can be averaged, or the actual values ​​of the candidate dimension items at multiple historical moments can be weighted summation to obtain the reference value of the candidate dimension items at the current moment. Compared with the related art that needs to predict the reference value of the candidate dimension items at the current moment based on the actual values ​​of the candidate dimension items at multiple historical moments, this solution is simple and easy to implement, and improves the efficiency of obtaining the reference value of the candidate dimension items at the current moment.

[0056] In some examples, the time interval between the historical moment and the current moment is less than 15 days. For example, the maximum time interval between the historical moment and the current moment is one week, 3 days, 1 day, etc. Compared with the related technology that needs to predict the reference value of the candidate dimension item at the current moment based on the actual value of the candidate dimension item at least 15 days before the current moment, this solution greatly reduces the amount of data and improves the efficiency of obtaining the reference value of the candidate dimension item at the current moment.

[0057] In some examples, before taking the weighted sum of the actual values ​​of the candidate dimension items at multiple historical moments, the weight of the actual value of the candidate dimension items at the historical moment is determined based on the time interval between the historical moment and the current moment. The weight is negatively correlated with the time interval, that is, the smaller the time interval, the greater the weight.

[0058] For example, the weights of the actual values ​​of the candidate dimension items at multiple historical moments may be determined based on a time decay algorithm. The time decay algorithm may include a geometric time decay algorithm. For example, if the current moment is August 10, the weight of the actual value of the candidate dimension item on August 9 is 1, the weight of the actual value of the candidate dimension item on August 8 is 1 / 2, and the weight of the actual value of the candidate dimension item on August 7 is 1 / 4.

[0059] It should be noted that the first dimension items are part of the candidate dimension items, and the number of the first dimension items is multiple. The single-dimensional root cause analysis result of the target indicator includes multiple first dimension items.

[0060] For example, taking the product as mobile phone and the target indicator as national sales amount, the candidate dimension items include the sales amount of mobile phones in province A, the sales amount of mobile phones in province B, the sales amount of mobile phones in province C, the sales amount of mobile phones in province D, the sales amount of mobile phones in sales channel 1, the sales amount of mobile phones in sales channel 2, and the sales amount of mobile phones in sales channel 3. The first dimension item (i.e. the unidimensional root cause analysis result of the national sales amount) includes the sales amount of mobile phones in province A, the sales amount of mobile phones in province B, and the sales amount of mobile phones in sales channel 1.

[0061] It should be noted that based on the data of multiple candidate dimension items, the first dimension item is screened out from multiple candidate dimension items, which can be implemented by using any single-dimensional root cause analysis algorithm in the relevant technology, and no excessive restrictions are made here. Among them, the single-dimensional root cause analysis algorithm may include an Adtributor algorithm. For example, the data of the candidate dimension items of the target indicator under multiple candidate dimensions can be input into the single-dimensional root cause analysis algorithm, and the single-dimensional root cause analysis algorithm outputs the first dimension item.

[0062] S103, based on the data of multiple candidate dimension items, a combination of second dimension items is screened out from the combination of multiple candidate dimension items as a multidimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of second dimension items belongs to a different candidate dimension.

[0063] It should be noted that each second dimension item in the combination of second dimension items belongs to a different candidate dimension. For example, if the combination of second dimension items includes the sales amount of mobile phones in Province A and sales channel 1, the second dimension items in the above combination of second dimension items include the sales amount of mobile phones in Province A and the sales amount of mobile phones in sales channel 1, among which the sales amount of mobile phones in Province A belongs to the administrative division candidate dimension, and the sales amount of mobile phones in sales channel 1 belongs to the sales channel candidate dimension.

[0064] It should be noted that there is no excessive limitation on the number of second dimensional items in the combination of second dimensional items, for example, it may include 2, 3, 4, etc.

[0065] It should be noted that the combination of second dimension items is a partial combination of candidate dimension items, and the number of the combination of second dimension items is multiple.

[0066] For example, taking the product as mobile phone and the target indicator as national sales amount, the candidate dimension items include the sales amount of mobile phones in province A, the sales amount of mobile phones in province B, the sales amount of mobile phones in province C, the sales amount of mobile phones in province D, the sales amount of mobile phones in sales channel 1, the sales amount of mobile phones in sales channel 2, and the sales amount of mobile phones in sales channel 3. The combination of the second dimension items (i.e., the multidimensional root cause analysis results of the national sales amount) includes the sales amount of mobile phones in province A and sales channel 1, and the sales amount of mobile phones in province B and sales channel 1.

[0067] It should be noted that based on the data of multiple candidate dimension items, the combination of the second dimension items can be screened out from the combination of multiple candidate dimension items, which can be achieved by using any multidimensional root cause analysis algorithm in the relevant technology, and no excessive restrictions are made here. Among them, the multidimensional root cause analysis algorithm may include Sqeeze, HotSpot, and RiskLoc algorithms. For example, the data of the candidate dimension items of the target indicators under multiple candidate dimensions can be input into the multidimensional root cause analysis algorithm, and the multidimensional root cause analysis algorithm outputs the combination of the second dimension items.

[0068] S104, obtaining a root cause analysis result of the target indicator based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator.

[0069] In one embodiment, based on the unidimensional root cause analysis result and the multidimensional root cause analysis result of the target indicator, a root cause analysis result of the target indicator is obtained, including using the unidimensional root cause analysis result and the multidimensional root cause analysis result of the target indicator as the root cause analysis result of the target indicator.

[0070] In one embodiment, based on the unidimensional root cause analysis results and multidimensional root cause analysis results of the target indicator, the root cause analysis results of the target indicator are obtained, including sorting the unidimensional root cause analysis results and multidimensional root cause analysis results of the target indicator according to the absolute value of the contribution, and taking the first Q results of the sorting as the root cause analysis results of the target indicator, wherein Q is a positive integer. It should be noted that the contribution acquisition process can be implemented by any contribution acquisition method in the relevant technology, and no excessive restrictions are made here. Therefore, in this method, Q results with larger absolute values ​​of contribution can be screened out from the unidimensional root cause analysis results and multidimensional root cause analysis results of the target indicator as the root cause analysis results of the target indicator.

[0071] In one embodiment, based on the unidimensional root cause analysis results and multidimensional root cause analysis results of the target indicator, the root cause analysis results of the target indicator are obtained, including normalizing the contribution of the unidimensional root cause analysis results and multidimensional root cause analysis results of the target indicator, sorting the unidimensional root cause analysis results and multidimensional root cause analysis results of the target indicator from large to small according to the absolute value of the contribution after normalization, and taking the first Q results of the sorting as the root cause analysis results of the target indicator, where Q is a positive integer. Therefore, in this method, the contribution can be normalized, and Q results with larger absolute values ​​of contribution after normalization can be screened out from the unidimensional root cause analysis results and multidimensional root cause analysis results of the target indicator as the root cause analysis results of the target indicator.

[0072] It should be noted that the normalization process can be implemented by any normalization process method in the relevant technology, and the contribution of the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator can be normalized together or separately, without too much limitation here.

[0073] In some examples, the contribution of the unidimensional root cause analysis result and the multidimensional root cause analysis result of the target indicator is normalized, including obtaining an overall quantitative parameter based on the contribution of the unidimensional root cause analysis result and the multidimensional root cause analysis result of the target indicator, and normalizing the contribution of the unidimensional root cause analysis result and the multidimensional root cause analysis result of the target indicator according to the overall quantitative parameter. In this embodiment, the normalization of the contribution of the unidimensional root cause analysis result of the target indicator is performed together with the normalization of the contribution of the multidimensional root cause analysis result of the target indicator.

[0074] In some examples, the contribution of the unidimensional root cause analysis result and the multidimensional root cause analysis result of the target indicator are normalized, including obtaining a first quantitative parameter based on the contribution of the unidimensional root cause analysis result of the target indicator, normalizing the contribution of the unidimensional root cause analysis result of the target indicator according to the first quantitative parameter, obtaining a second quantitative parameter based on the contribution of the multidimensional root cause analysis result of the target indicator, and normalizing the contribution of the multidimensional root cause analysis result of the target indicator according to the second quantitative parameter. In this embodiment, the normalization of the contribution of the unidimensional root cause analysis result of the target indicator and the normalization of the contribution of the multidimensional root cause analysis result of the target indicator are performed separately.

[0075] In some examples, the contribution of the multidimensional root cause analysis results of the target indicator is normalized, including splitting the combination of second dimension items to obtain multiple second dimension items, updating the contribution of the multidimensional root cause analysis results of the target indicator based on the number of candidate dimensions to which the multiple second dimension items belong, and normalizing the updated contribution of the multidimensional root cause analysis results of the target indicator.

[0076] For example, if the combination of second dimension items (i.e., the multidimensional root cause analysis results of the national sales amount) includes the sales amount of mobile phones in province A and sales channel 1, and the sales amount of mobile phones in province B and sales channel 1, the sales amount of mobile phones in province A and sales channel 1 can be split to obtain the sales amount of mobile phones in province A and the sales amount of mobile phones in sales channel 1, and the sales amount of mobile phones in province B and sales channel 1 can be split to obtain the sales amount of mobile phones in province B and the sales amount of mobile phones in sales channel 1. Then, multiple second dimension items include the sales amount of mobile phones in province A, the sales amount of mobile phones in province B, and the sales amount of mobile phones in sales channel 1, and the candidate dimensions to which the multiple second dimension items belong include administrative divisions and sales channels, that is, the number of candidate dimensions to which the multiple second dimension items belong is 2.

[0077] For example, based on the number of candidate dimensions to which multiple second dimension items belong, the contribution of the multidimensional root cause analysis result of the target indicator is updated, including obtaining the product of the contribution of the multidimensional root cause analysis result of the target indicator and the number of candidate dimensions to which multiple second dimension items belong, and updating the contribution of the multidimensional root cause analysis result of the target indicator based on the product. It can be understood that the contribution of the single-dimensional root cause analysis result of the target indicator is greater than the contribution of the multidimensional root cause analysis result of the target indicator, and the number of candidate dimensions to which multiple second dimension items belong is greater than 1, so the contribution of the multidimensional root cause analysis result of the target indicator can be amplified, thereby reducing the difference between the contribution of the single-dimensional root cause analysis result of the target indicator and the contribution of the multidimensional root cause analysis result of the target indicator, making the sorting result more accurate, thereby improving the accuracy of the root cause analysis.

[0078] The root cause analysis method provided by the embodiment of the present disclosure obtains the target indicator of the product and the candidate dimension items of the target indicator under multiple candidate dimensions, wherein all the candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator, and based on the data of the multiple candidate dimension items, the first dimension item is screened out from the multiple candidate dimension items as the single-dimensional root cause analysis result of the target indicator, and based on the data of the multiple candidate dimension items, the combination of the second dimension items is screened out from the combination of the multiple candidate dimension items as the multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension, and the root cause analysis result of the target indicator is obtained based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator. Thus, the target indicator can be subjected to single-dimensional root cause analysis to obtain the first dimension item, and the target indicator can also be subjected to multi-dimensional root cause analysis to obtain the combination of the second dimension item, and the root cause analysis result of the target indicator is obtained by comprehensively considering the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result, thereby improving the comprehensiveness of the root cause analysis.

[0079] Figure 2 is a flow chart of a root cause analysis method according to another exemplary embodiment. Figure 2 As shown, the root cause analysis method of the embodiment of the present disclosure includes the following steps.

[0080] S201, obtaining a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator.

[0081] For the relevant contents of step S201, please refer to the above embodiment, which will not be described again here.

[0082] S202: Determine a sixth dimension to be deleted from multiple candidate dimensions based on at least one of the number of candidate dimension items under the candidate dimension, the name of the candidate dimension, and the user's preference for the candidate dimension.

[0083] It should be noted that the sixth dimension is a partial candidate dimension, and the number of the sixth dimension is at least one.

[0084] In one embodiment, obtaining the user's preference for the candidate dimension includes obtaining the user's preference for the candidate dimension based on the user's user profile. For example, the user's user profile can be obtained based on the user's historical behavior associated with the root cause analysis. The user refers to the manufacturer, seller, etc. of the product.

[0085] In one implementation, based on at least one of the number of candidate dimension items under the candidate dimension, the name of the candidate dimension, and the user's preference for the candidate dimension, the sixth dimension to be deleted is determined from multiple candidate dimensions, including the following possible implementations:

[0086] Method 1: If the number of candidate dimension items under the candidate dimension is greater than or equal to the first set threshold, or the number of candidate dimension items under the candidate dimension is less than or equal to the second set threshold, the candidate dimension is determined as the sixth dimension.

[0087] It should be noted that there are no excessive restrictions on the first set threshold and the second set threshold. For example, the first set threshold is 10,000 and the second set threshold is 2.

[0088] Therefore, in this method, when the number of candidate dimension items under the candidate dimension is too large, the candidate dimension can be determined as the sixth dimension, which can greatly reduce the amount of data to be processed. When the number of candidate dimension items under the candidate dimension is too small, the candidate dimension can be determined as the sixth dimension.

[0089] Method 2: If the name of the candidate dimension is a set name, the candidate dimension is determined as the sixth dimension.

[0090] It should be noted that there are not too many restrictions on the setting name, for example, it can include an identifier, a type, etc.

[0091] Method 3: If the user's preference for the candidate dimension is less than or equal to the third set threshold, the candidate dimension is determined as the sixth dimension.

[0092] It should be noted that there are no excessive restrictions on the third set threshold. For example, if the value range of the preference is 0-10, the third set threshold may be 4.

[0093] Therefore, in this method, when the user's preference for the candidate dimension is relatively small, the candidate dimension can be determined as the sixth dimension.

[0094] S203: Delete the sixth dimension from the multiple candidate dimensions, and determine the multiple candidate dimensions remaining after the deletion as the first dimension.

[0095] S204, based on the data of the candidate dimension items under the multiple first dimensions, screen out the first dimension items from the candidate dimension items under the multiple first dimensions as the single-dimensional root cause analysis result of the target indicator.

[0096] For example, if multiple candidate dimensions include candidate dimensions 1 to 10, and the sixth dimension includes candidate dimensions 1 to 4, candidate dimensions 1 to 4 can be deleted from candidate dimensions 1 to 10, and candidate dimensions 5 to 10 remaining after the deletion can be determined as the first dimension, and based on the data of the candidate dimension items under candidate dimensions 5 to 10, the first dimension items can be screened out from the candidate dimension items under candidate dimensions 5 to 10 as the single-dimensional root cause analysis results of the target indicator.

[0097] S205, based on the data of multiple candidate dimension items, a combination of second dimension items is screened out from the combination of multiple candidate dimension items as the multidimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of second dimension items belongs to a different candidate dimension.

[0098] S206, obtaining a root cause analysis result of the target indicator based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator.

[0099] For the relevant contents of steps S205-S206, please refer to the above embodiment and will not be repeated here.

[0100] The root cause analysis method provided by the embodiment of the present disclosure determines the sixth dimension to be deleted from multiple candidate dimensions based on the number of candidate dimension items under the candidate dimensions, the name of the candidate dimensions, and at least one of the user's preference for the candidate dimensions, deletes the sixth dimension from the multiple candidate dimensions, and determines the multiple candidate dimensions remaining after the deletion as the first dimension, and screens out the first dimension items from the multiple candidate dimension items under the first dimensions based on the data of the candidate dimension items under the multiple first dimensions. Thus, the sixth dimension can be determined by considering at least one of the number of candidate dimension items under the candidate dimensions, the name of the candidate dimensions, and the user's preference for the candidate dimensions, and the sixth dimension can be deleted from the multiple candidate dimensions to obtain the first dimension, that is, the number of first dimensions is small, and the single-dimensional root cause analysis result is obtained based on the data of the candidate dimension items under the multiple first dimensions, which can greatly reduce the amount of data to be processed in the single-dimensional root cause analysis and improve the efficiency of the single-dimensional root cause analysis.

[0101] Figure 3 is a flow chart of a root cause analysis method according to another exemplary embodiment. Figure 3 As shown, the root cause analysis method of the embodiment of the present disclosure includes the following steps.

[0102] S301, obtaining a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator.

[0103] For the relevant contents of step S301, please refer to the above embodiment, which will not be repeated here.

[0104] S302, based on the data of the candidate dimension items, obtaining the explanatory power and surprise degree of the candidate dimension items.

[0105] It should be noted that EP (Explanatory Power) is used to characterize the proportion of the change value of the target indicator caused by the candidate dimension item. For example, it can be obtained based on the ratio of the change value of the candidate dimension item to the change value of the target indicator. Surprise is used to characterize the deviation between the actual value of the candidate dimension item and the reference value. The greater the deviation between the actual value of the candidate dimension item and the reference value, the greater the surprise, and the more likely it is to be the root cause of the abnormality of the target indicator.

[0106] The process of acquiring explanatory power can be implemented by using any method for acquiring explanatory power in the relevant technology, and the process of acquiring surprise degree can be implemented by using any method for acquiring surprise degree in the relevant technology, and no further limitations are made here.

[0107] In one implementation, based on the data of the candidate dimension items, the specific process of obtaining the explanatory power and surprise degree of the candidate dimension items is as follows:

[0108] EP ij (m)=(A ij (m)-F ij (m)0 / (A(m)-F(m))

[0109] P ij (m) = F ij (m) / F(m)

[0110] q ij (m) = A ij (m) / A(m)

[0111] S ij (m) = JSD(P ij (m)||q ij (m)0

[0112] Among them, EP ij (m) is the explanatory power of the jth candidate dimension item under the i-th candidate dimension, A ij (m) is the actual value of the jth candidate dimension item under the i-th candidate dimension at the current moment, F ij (m) is the reference value of the jth candidate dimension item under the i-th candidate dimension at the current moment, A(m) is the actual value of the target indicator at the current moment, F(m) is the reference value of the target indicator at the current moment, S ij (m) is the surprise of the j-th candidate dimension item under the i-th candidate dimension, and JSD(·) is the operator symbol of JS (Jensen-Shannon) divergence.

[0113] S303, sort multiple candidate dimension items from large to small according to the absolute value of explanatory power, add the top N candidate dimension items to the target set, and / or add the candidate dimension items whose absolute value of explanatory power is greater than a fourth set threshold to the target set, where N is a positive integer.

[0114] S304, sorting the multiple candidate dimension items in the target set from large to small according to the absolute value of the surprise degree, and determining the first M candidate dimension items as the first dimension items as the single-dimensional root cause analysis result of the target indicator, where M is a positive integer.

[0115] It should be noted that there is no excessive limitation on the fourth set threshold value. M≤N, for example, N=20, M=6.

[0116] For example, if multiple candidate dimension items include candidate dimension items 1 to 100, candidate dimension items 1 to 100 can be sorted from large to small according to the absolute value of explanatory power, and the top 20 candidate dimension items can be added to the target set; if multiple candidate dimension items in the target set include candidate dimension items 1 to 20, candidate dimension items 1 to 20 can be sorted from large to small according to the absolute value of surprise, and the top 6 candidate dimension items can be determined as the first dimension items.

[0117] For example, if multiple candidate dimension items include candidate dimension items 1 to 100, if the absolute values ​​of the explanatory power of candidate dimension items 5 to 24 are greater than the fourth set threshold, and the absolute values ​​of the explanatory power of candidate dimension items 1-4 and 25 to 100 are all less than or equal to the fourth set threshold, candidate dimension items 5 to 24 are added to the target set, and candidate dimension items 5 to 24 are sorted from large to small according to the absolute value of the surprise degree, and the top 6 candidate dimension items are determined as the first dimension items.

[0118] S305, based on the data of multiple candidate dimension items, a combination of second dimension items is screened out from the combination of multiple candidate dimension items as the multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of second dimension items belongs to a different candidate dimension.

[0119] S306, obtaining a root cause analysis result of the target indicator based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator.

[0120] For the relevant contents of steps S305-S306, please refer to the above embodiment and will not be repeated here.

[0121] The root cause analysis method provided by the embodiment of the present disclosure obtains the explanatory power and surprise degree of the candidate dimension items based on the data of the candidate dimension items, sorts the multiple candidate dimension items from large to small according to the absolute value of the explanatory power, adds the first N candidate dimension items to the target set, and / or, adds the candidate dimension items whose absolute value of the explanatory power is greater than the fourth set threshold to the target set, wherein N is a positive integer, sorts the multiple candidate dimension items in the target set from large to small according to the absolute value of the surprise degree, and determines the first M candidate dimension items as the first dimension items as the single-dimensional root cause analysis result of the target indicator, wherein M is a positive integer. Thus, the candidate dimension items with larger absolute values ​​of explanatory power can be screened out from the multiple candidate dimension items and added to the target set. Compared with the problem of generating the target set by sequentially traversing the candidate dimension items in the related art, which easily ignores the candidate dimension items with larger absolute values ​​of explanatory power at the back, the candidate dimension items with larger absolute values ​​of explanatory power will not be omitted in this solution, thereby improving the accuracy of the single-dimensional root cause analysis.

[0122] Figure 4 is a flow chart of a root cause analysis method according to another exemplary embodiment. Figure 4 As shown, the root cause analysis method of the embodiment of the present disclosure includes the following steps.

[0123] S401, obtaining a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator.

[0124] For the relevant contents of step S401, please refer to the above embodiment, which will not be repeated here.

[0125] S402, based on the actual value and reference value of the candidate dimension item at the current moment, and the actual value and reference value of the target indicator at the current moment, obtain the contribution of the candidate dimension item.

[0126] It should be noted that the contribution degree refers to the contribution of the change value of the candidate dimension item to the change value of the target indicator. The contribution degree acquisition process can be implemented by any contribution degree acquisition method in the relevant technology, and no excessive restrictions are made here.

[0127] In one implementation, the process of obtaining the contribution of the candidate dimension items is as follows:

[0128] B ij (m)=(A ij (m)-F ij (m)) / (A(m)-F(m))

[0129] Among them, B ij(m) is the contribution of the jth candidate dimension item under the i-th candidate dimension.

[0130] In one implementation, the process of obtaining the contribution of the candidate dimension items is as follows:

[0131] B ij (m)=(A ij (m)-A ij (mx)) / (A(m)-A(mx))

[0132] Among them, A ij (mx) is the actual value of the jth candidate dimension item under the i-th candidate dimension at the historical moment, and A(mx) is the actual value of the target indicator at the historical moment. x is the time interval between the historical moment and the current moment. For example, if the current moment is August 10, the granularity of the time interval is days, and A ij (m) is the actual value of the jth candidate dimension item under the i-th candidate dimension on August 10, A ij (m-1) is the actual value of the jth candidate dimension item under the i-th candidate dimension on August 9, A ij (m-2) is the actual value of the j-th candidate dimension item under the i-th candidate dimension on August 8.

[0133] S403, sorting multiple candidate dimension items from large to small according to the absolute value of contribution, determining the first M candidate dimension items as the first dimension items as the single-dimensional root cause analysis result of the target indicator, where M is a positive integer.

[0134] For example, if multiple candidate dimension items include candidate dimension items 1 to 100, candidate dimension items 1 to 100 may be sorted from large to small according to the absolute value of contribution, and the first six candidate dimension items may be determined as first dimension items.

[0135] S404, based on the data of multiple candidate dimension items, a combination of second dimension items is screened out from the combination of multiple candidate dimension items as a multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of second dimension items belongs to a different candidate dimension.

[0136] S405, obtaining a root cause analysis result of the target indicator based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator.

[0137] For the relevant contents of steps S404-S405, please refer to the above embodiment and will not be repeated here.

[0138] The root cause analysis method provided by the embodiment of the present disclosure obtains the contribution of the candidate dimension items based on the actual value and reference value of the candidate dimension items at the current moment, and the actual value and reference value of the target indicator at the current moment, and sorts the multiple candidate dimension items from large to small according to the absolute value of the contribution, and determines the first M candidate dimension items in the sorting as the first dimension items as the single-dimensional root cause analysis result of the target indicator, where M is a positive integer. In this way, the contribution of the candidate dimension items can be directly obtained, and the candidate dimension items with larger absolute values ​​of contribution can be screened out from the multiple candidate dimension items as the first dimension items, which is simple and easy to implement, and improves the efficiency of the single-dimensional root cause analysis.

[0139] Figure 5 is a flow chart of a root cause analysis method according to another exemplary embodiment. Figure 5 As shown, the root cause analysis method of the embodiment of the present disclosure includes the following steps.

[0140] S501, obtaining a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator.

[0141] S502, based on the data of the plurality of candidate dimension items, a first dimension item is screened out from the plurality of candidate dimension items as a single-dimensional root cause analysis result of the target indicator.

[0142] For the relevant contents of steps S501-S502, please refer to the above embodiment and will not be repeated here.

[0143] S503, based on the data of multiple candidate dimension items, filter out the combination of third dimension items of the i-th round from the combination of multiple candidate dimension items, where i is a positive integer and the initial value of i is 1.

[0144] It should be noted that the combinations of third dimensional items in different rounds may be the same or different. For example, the numbers of third dimensional items in the combinations of third dimensional items in different rounds are different.

[0145] In one embodiment, the number of combinations of third-dimensional items in the i-th round is negatively correlated with i, that is, the number of combinations of third-dimensional items decreases as the rounds increase.

[0146] In one embodiment, the number of third dimensional items in the combination of third dimensional items in the i-th round is i+1, that is, the i-th round is an i+1-dimensional root cause analysis, for example, the first round is a two-dimensional root cause analysis, the second round is a three-dimensional root cause analysis, the third round is a four-dimensional root cause analysis, and so on.

[0147] In one implementation, based on data of multiple candidate dimension items, a combination of third dimension items of the i-th round is screened out from the combination of multiple candidate dimension items, including the following possible implementations:

[0148] Method 1: if i=1, based on the data of the first dimension items under any two candidate dimensions, the combination of the third dimension items of the first round is screened out from the combination of the first dimension items under any two candidate dimensions.

[0149] It can be understood that the first dimension item is a single-dimensional root cause analysis result, and the combination of the third dimension items in the first round is a two-dimensional root cause analysis result. Therefore, in this method, a two-dimensional root cause analysis can be performed based on the single-dimensional root cause analysis result.

[0150] For example, the candidate dimensions include dimensions A, B, and C. Based on the data of the first dimension items under dimensions A and B, the data of the first dimension items under dimensions A and C, and the data of the first dimension items under dimensions B and C, the combination of the first dimension items under dimensions A and B, the combination of the first dimension items under dimensions A and C, and the combination of the first dimension items under dimensions B and C can be used to screen out the combination of the third dimension items for the first round.

[0151] Among them, if the first dimension items under dimension A include dimension items A1 and A2, and the first dimension items under dimension B include dimension items B1 and B2, then the combination of the first dimension items under dimensions A and B includes the combination of A1 B1 (that is, the combination of dimension items A1 and B1), the combination of A1 B2 (that is, the combination of dimension items A1 and B2), the combination of A2 B1 (that is, the combination of dimension items A2 and B1), and the combination of A2 B2 (that is, the combination of dimension items A2 and B2).

[0152] It should be noted that the combination of the first dimension items under dimensions A and C and the combination of the first dimension items under dimensions B and C can refer to the combination of the first dimension items under dimensions A and B, and will not be repeated here.

[0153] In some examples, based on the data of the first dimension items under any two candidate dimensions, a combination of third dimension items of the first round is screened out from the combination of the first dimension items under any two candidate dimensions, including associating the data of the first dimension items under any two candidate dimensions to obtain the associated data of the combination of the first dimension items, and based on the associated data of the combination of multiple first dimension items, a combination of third dimension items of the first round is screened out from the combination of the first dimension items under any two candidate dimensions.

[0154] For example, if the first dimension items include the sales amount of mobile phones in province A, the sales amount of mobile phones in sales channel 1, and the sales amount of mobile phones in sales channel 2, the data of the sales amount of mobile phones in province A and the sales amount of mobile phones in sales channel 1 can be associated to obtain the associated data of the combination of the first dimension items. The data of the sales amount of mobile phones in province A and the sales amount of mobile phones in sales channel 2 can also be associated to obtain the associated data of the combination of the first dimension items.

[0155] It should be noted that based on the associated data of the combination of multiple first dimension items, the combination of third dimension items of the first round is screened out from the combination of first dimension items under any two candidate dimensions. Please refer to the relevant content of step S102, which will not be repeated here.

[0156] In some examples, based on the associated data of the combination of multiple first dimension items, the combination of third dimension items of the first round is screened out from the combination of first dimension items under any two candidate dimensions, including inputting the associated data of the combination of multiple first dimension items into a single-dimensional root cause analysis algorithm, and obtaining the combination of third dimension items of the first round by the single-dimensional root cause analysis algorithm.

[0157] Method 2: If i ≥ 2, split the combination of the third dimension items of the i-1th round to obtain multiple third dimension items of the i-1th round, use the multiple third dimension items of the i-1th round as the fifth dimension items, and based on the data of the fifth dimension items under any i+1 candidate dimensions, screen out the combination of the third dimension items of the i-1th round from the combination of the fifth dimension items under any i+1 candidate dimensions.

[0158] It can be understood that the combination of the third dimension items in the i-1th round is the i-1th dimension root cause analysis result, and the combination of the third dimension items in the i-th round is the i-th dimension root cause analysis result. Therefore, in this method, the i-th dimension root cause analysis can be performed based on the i-1th dimension root cause analysis result.

[0159] For example, taking i=2 as an example, the combination of the third dimension items of the first round can be split to obtain multiple third dimension items of the first round. For example, if the combination of the third dimension items of the first round is the sales amount of mobile phones in province A and sales channel 1, the sales amount of mobile phones in province A and sales channel 1 can be split to obtain the sales amount of mobile phones in province A and the sales amount of mobile phones in sales channel 1, that is, the multiple third dimension items of the first round include the sales amount of mobile phones in province A and the sales amount of mobile phones in sales channel 1.

[0160] For example, taking i=2, the candidate dimensions include dimensions A, B, and C. Based on the data of the fifth dimension items under dimensions A, B, and C, the combination of the third dimension items for the second round can be screened out from the combination of the fifth dimension items under dimensions A, B, and C.

[0161] Among them, if the fifth dimension items under dimension A include dimension items A1 and A2, the fifth dimension items under dimension B include dimension item B1, and the fifth dimension items under dimension C include dimension item C2, then the combination of the fifth dimension items under dimensions A, B, and C includes the combination of A1 B1C2 (that is, the combination of dimension items A1, B1, and C2) and the combination of A2 B1 C2 (that is, the combination of dimension items A2, B1, and C2).

[0162] In some examples, based on the data of the fifth dimension items under any i+1 candidate dimensions, combinations of third dimension items of the i-th round are screened out from combinations of the fifth dimension items under any i+1 candidate dimensions, including associating the data of the fifth dimension items under any i+1 candidate dimensions to obtain associated data of the combinations of fifth dimension items, and based on the associated data of multiple combinations of fifth dimension items, combinations of third dimension items of the i-th round are screened out from combinations of fifth dimension items under any i+1 candidate dimensions.

[0163] For instance, taking i=2 as an example, if the fifth dimension items include the sales amount of mobile phones in province A, the sales amount of mobile phones in sales channel 1, the sales amount of mobile phones in China, and the sales amount of mobile phones abroad, the data of the sales amount of mobile phones in province A, the sales amount of mobile phones in sales channel 1, and the sales amount of mobile phones in China can be associated to obtain the associated data of the combination of items of the fifth dimension. The data of the sales amount of mobile phones in province A, the sales amount of mobile phones in sales channel 1, and the sales amount of mobile phones abroad can also be associated to obtain the associated data of the combination of items of the fifth dimension.

[0164] It should be noted that based on the associated data of the combination of multiple fifth dimension items, the combination of third dimension items of the i-th round is screened out from the combination of fifth dimension items under any i+1 candidate dimensions. Please refer to the relevant content of step S102, which will not be repeated here.

[0165] In some examples, based on the associated data of multiple combinations of fifth dimension items, combinations of third dimension items of the i-th round are screened out from the combinations of fifth dimension items under any i+1 candidate dimensions, including inputting the associated data of multiple combinations of fifth dimension items into a single-dimensional root cause analysis algorithm, and obtaining the combinations of third dimension items of the i-th round by the single-dimensional root cause analysis algorithm.

[0166] S504: Determine whether the number of combinations of third dimension items in the i-th round is less than or equal to a fifth set threshold.

[0167] If yes, that is, the number of combinations of third dimensional items in the i-th round is less than or equal to the fifth set threshold, execute step S505; if no, that is, the number of combinations of third dimensional items in the i-th round is greater than the fifth set threshold, execute step S506.

[0168] S505, taking the combination of the third dimension items of the i-th round as the combination of the fourth dimension items, and based on the combination of the fourth dimension items, obtaining the combination of the second dimension items as the multi-dimensional root cause analysis result of the target indicator.

[0169] S506, continue to execute the screening process of the i+1th round until the number of combinations of the third dimension items of the target round is less than or equal to the fifth set threshold, use the combination of the third dimension items of the target round as the combination of the fourth dimension items, and based on the combination of the fourth dimension items, obtain the combination of the second dimension items as the multi-dimensional root cause analysis result of the target indicator.

[0170] It should be noted that there are no excessive restrictions on the fifth set threshold, for example, it can be 100,000.

[0171] For instance, taking i=1 as an example, if the number of combinations of third-dimensional items in the first round is greater than the fifth set threshold, it indicates that the number of combinations of third-dimensional items in the first round is relatively large, and the screening process of the second round can be continued; if the number of combinations of third-dimensional items in the second round is greater than the fifth set threshold, it indicates that the number of combinations of third-dimensional items in the second round is relatively large, and the screening process of the third round can be continued; if the number of combinations of third-dimensional items in the third round is less than or equal to the fifth set threshold, it indicates that the number of combinations of third-dimensional items in the third round is relatively small, that is, the third round is the target round, and the combinations of third-dimensional items in the third round can be used as combinations of fourth-dimensional items, and based on the combinations of third-dimensional items in the third round, the combinations of second-dimensional items can be obtained.

[0172] In one embodiment, based on the combination of fourth dimension items, a combination of second dimension items is obtained, including splitting the combination of fourth dimension items to obtain multiple fourth dimension items, and based on the data of the multiple fourth dimension items, screening out the combination of second dimension items from the combination of the multiple fourth dimension items.

[0173] It should be noted that based on the data of multiple fourth dimensional items, a combination of second dimensional items is screened out from the combination of multiple fourth dimensional items. The relevant content of step S103 in the above embodiment can be referred to and will not be repeated here.

[0174] In some examples, based on data of multiple fourth dimension items, combinations of second dimension items are screened out from combinations of multiple fourth dimension items, including inputting data of multiple fourth dimension items into a multidimensional root cause analysis algorithm, and obtaining combinations of second dimension items from the multidimensional root cause analysis algorithm.

[0175] S507, obtaining a root cause analysis result of the target indicator based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator.

[0176] For the relevant contents of step S507, please refer to the above embodiment, which will not be repeated here.

[0177] The root cause analysis method provided by the embodiment of the present disclosure is based on the data of multiple candidate dimension items, and from the combination of multiple candidate dimension items, the combination of the third dimension items of the i-th round is screened out, and it is determined whether the number of the combination of the third dimension items of the i-th round is less than or equal to the fifth set threshold. If the number of the combination of the third dimension items of the i-th round is less than or equal to the fifth set threshold, the combination of the third dimension items of the i-th round is used as the combination of the fourth dimension items, and based on the combination of the fourth dimension items, the combination of the second dimension items is obtained. If the number of the combination of the third dimension items of the i-th round is greater than the fifth set threshold, the screening process of the i+1th round is continued until the number of the combination of the third dimension items of the target round is less than or equal to the fifth set threshold, the combination of the third dimension items of the target round is used as the combination of the fourth dimension items, and based on the combination of the fourth dimension items, the combination of the second dimension items is obtained. Thus, multiple rounds of screening processes can be performed until the number of the combination of the third dimension items of the target round is small, that is, the number of the combination of the fourth dimension items is small, and the multi-dimensional root cause analysis results are obtained based on the combination of the fourth dimension items, which greatly reduces the amount of data to be processed in the multi-dimensional root cause analysis and improves the efficiency of the multi-dimensional root cause analysis.

[0178] Figure 6 FIG. 1 is a block diagram of a root cause analysis device according to an exemplary embodiment. Figure 6 The root cause analysis device 100 of the embodiment of the present disclosure includes: a first acquisition module 110, a first screening module 120, a second screening module 130 and a second acquisition module 140.

[0179] The first acquisition module 110 is configured to execute acquisition of a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator;

[0180] The first screening module 120 is configured to perform data based on the plurality of candidate dimension items, and screen out a first dimension item from the plurality of candidate dimension items as a single-dimensional root cause analysis result of the target indicator;

[0181] The second screening module 130 is configured to perform, based on the data of the plurality of candidate dimension items, screening out a combination of second dimension items from the combination of the plurality of candidate dimension items as a multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension;

[0182] The second acquisition module 140 is configured to execute a single-dimensional root cause analysis result and a multi-dimensional root cause analysis result based on the target indicator to obtain a root cause analysis result of the target indicator.

[0183] In one embodiment of the present disclosure, the first screening module 120 is further configured to perform: determining a sixth dimension to be deleted from a plurality of the candidate dimensions based on at least one of the number of candidate dimension items under the candidate dimensions, the name of the candidate dimensions, and the user's preference for the candidate dimensions; deleting the sixth dimension from the plurality of the candidate dimensions, and determining the plurality of candidate dimensions remaining after the deletion as the first dimension; and screening out the first dimension items from the plurality of candidate dimension items under the first dimension based on the data of the plurality of candidate dimension items under the first dimension.

[0184] In one embodiment of the present disclosure, the first screening module 120 is further configured to execute: if the number of candidate dimension items under the candidate dimension is greater than or equal to a first set threshold, or the number of candidate dimension items under the candidate dimension is less than or equal to a second set threshold, the candidate dimension is determined as the sixth dimension; and / or, if the name of the candidate dimension is a set name, the candidate dimension is determined as the sixth dimension; and / or, if the user's preference for the candidate dimension is less than or equal to a third set threshold, the candidate dimension is determined as the sixth dimension.

[0185] In one embodiment of the present disclosure, the first screening module 120 is further configured to perform: based on the data of the candidate dimension items, obtaining the explanatory power and surprise degree of the candidate dimension items; sorting the multiple candidate dimension items from large to small according to the absolute value of the explanatory power, adding the top N candidate dimension items to the target set, and / or adding the candidate dimension items whose absolute value of the explanatory power is greater than a fourth set threshold to the target set, wherein N is a positive integer; sorting the multiple candidate dimension items in the target set from large to small according to the absolute value of the surprise degree, and determining the top M candidate dimension items in the target set as the first dimension items, wherein M is a positive integer.

[0186] In one embodiment of the present disclosure, the first screening module 120 is further configured to perform: obtaining the contribution of the candidate dimension item based on the actual value and reference value of the candidate dimension item at the current moment, and the actual value and reference value of the target indicator at the current moment; sorting the multiple candidate dimension items from large to small according to the absolute value of the contribution, and determining the first M candidate dimension items in the sort as the first dimension items, where M is a positive integer.

[0187] In one embodiment of the present disclosure, the second screening module 130 is further configured to perform: based on the data of the plurality of candidate dimension items, screening out the combination of the third dimension items of the i-th round from the combination of the plurality of candidate dimension items, wherein i is a positive integer and the initial value of i is 1; judging whether the number of the combination of the third dimension items of the i-th round is less than or equal to the fifth set threshold; if the number of the combination of the third dimension items of the i-th round is less than or equal to the fifth set threshold, taking the combination of the third dimension items of the i-th round as the combination of the fourth dimension items, and obtaining the combination of the second dimension items based on the combination of the fourth dimension items; if the number of the combination of the third dimension items of the i-th round is greater than the fifth set threshold, continuing to perform the screening process of the i+1th round until the number of the combination of the third dimension items of the target round is less than or equal to the fifth set threshold, taking the combination of the third dimension items of the target round as the combination of the fourth dimension items, and obtaining the combination of the second dimension items based on the combination of the fourth dimension items.

[0188] In one embodiment of the present disclosure, the second screening module 130 is further configured to execute: if i=1, based on the data of the first dimension items under any two candidate dimensions, a combination of the third dimension items of the first round is screened out from the combination of the first dimension items under any two candidate dimensions.

[0189] In one embodiment of the present disclosure, the second screening module 130 is further configured to perform: if i≥2, split the combination of the third dimension items of the i-1th round to obtain multiple third dimension items of the i-1th round; use the multiple third dimension items of the i-1th round as fifth dimension items; based on the data of the fifth dimension items under any i+1 candidate dimensions, screen out the combination of the third dimension items of the i-1th round from the combination of the fifth dimension items under any i+1 candidate dimensions.

[0190] In one embodiment of the present disclosure, the second screening module 130 is further configured to execute: splitting the combination of the fourth dimension items to obtain a plurality of fourth dimension items; and screening out the combination of the second dimension items from the combination of the plurality of fourth dimension items based on the data of the plurality of fourth dimension items.

[0191] In one embodiment of the present disclosure, the data of the candidate dimension item includes a reference value of the candidate dimension item at the current moment, and the first acquisition module 110 is further configured to execute: obtaining actual values ​​of the candidate dimension item at multiple historical moments; taking the average of the actual values ​​of the candidate dimension item at multiple historical moments as the reference value of the candidate dimension item at the current moment; or, performing weighted summation of the actual values ​​of the candidate dimension item at multiple historical moments to obtain the reference value of the candidate dimension item at the current moment.

[0192] In one embodiment of the present disclosure, the second acquisition module 140 is further configured to execute: normalizing the contribution of the single-dimensional root cause analysis results and the multi-dimensional root cause analysis results of the target indicator; sorting the single-dimensional root cause analysis results and the multi-dimensional root cause analysis results of the target indicator from large to small according to the absolute value of the contribution after normalization, and taking the first Q results of the sorting as the root cause analysis results of the target indicator, where Q is a positive integer.

[0193] In one embodiment of the present disclosure, the second acquisition module 140 is further configured to execute: splitting the combination of the second dimensional items to obtain a plurality of second dimensional items; updating the contribution of the multidimensional root cause analysis result of the target indicator based on the number of candidate dimensions to which the plurality of second dimensional items belong; and normalizing the updated contribution of the multidimensional root cause analysis result of the target indicator.

[0194] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0195] The root cause analysis device provided by the embodiment of the present disclosure obtains the target indicator of the product and the candidate dimension items of the target indicator under multiple candidate dimensions, wherein all the candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator, and based on the data of the multiple candidate dimension items, the first dimension item is screened out from the multiple candidate dimension items as the single-dimensional root cause analysis result of the target indicator, and based on the data of the multiple candidate dimension items, the combination of the second dimension items is screened out from the combination of the multiple candidate dimension items as the multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension, and the root cause analysis result of the target indicator is obtained based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator. Thus, the target indicator can be subjected to single-dimensional root cause analysis to obtain the first dimension item, and the target indicator can also be subjected to multi-dimensional root cause analysis to obtain the combination of the second dimension item, and the root cause analysis result of the target indicator is obtained by comprehensively considering the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result, thereby improving the comprehensiveness of the root cause analysis.

[0196] Figure 7 is a block diagram of an electronic device 200 according to an exemplary embodiment.

[0197] like Figure 7 As shown, the electronic device 200 includes:

[0198] The memory 210 and the processor 220, and the bus 230 connecting different components (including the memory 210 and the processor 220), the memory 210 stores a computer program, and when the processor 220 executes the program, the root cause analysis method described in the embodiment of the present disclosure is implemented.

[0199] Bus 230 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. For example, these architectures include but are not limited to Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus and Peripheral Component Interconnect (PCI) bus.

[0200] The electronic device 200 typically includes a variety of electronic device readable media. These media can be any available media that can be accessed by the electronic device 200, including volatile and non-volatile media, removable and non-removable media.

[0201] The memory 210 may also include computer system readable media in the form of volatile memory, such as random access memory (RAM) 240 and / or cache memory 250. The electronic device 200 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 260 may be used to read and write non-removable, non-volatile magnetic media ( Figure 7 not shown, usually called a "hard drive"). Although Figure 7 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 230 via one or more data medium interfaces. The memory 210 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present disclosure.

[0202] A program / utility 280 having a set (at least one) of program modules 270 may be stored, for example, in the memory 210, such program modules 270 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 270 generally perform the functions and / or methods of the embodiments described in the present disclosure.

[0203] The electronic device 200 may also communicate with one or more external devices 290 (e.g., keyboards, pointing devices, displays 291, etc.), one or more devices that enable a user to interact with the electronic device 200, and / or any device that enables the electronic device 200 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed through an input / output (I / O) interface 292. Furthermore, the electronic device 200 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter 293. Figure 7 As shown, the network adapter 293 communicates with other modules of the electronic device 200 via the bus 230. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0204] The processor 220 executes various functional applications and data processing by running the programs stored in the memory 210 .

[0205] It should be noted that the implementation process and technical principles of the electronic device of this embodiment refer to the aforementioned explanation of the root cause analysis method of the embodiment of the present disclosure, and will not be repeated here.

[0206] The electronic device provided by the embodiment of the present disclosure can execute the root cause analysis method as described above, obtain the target indicator of the product, and the candidate dimension items of the target indicator under multiple candidate dimensions, wherein all the candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator, and based on the data of the multiple candidate dimension items, the first dimension item is screened out from the multiple candidate dimension items as the single-dimensional root cause analysis result of the target indicator, and based on the data of the multiple candidate dimension items, the combination of the second dimension items is screened out from the combination of the multiple candidate dimension items as the multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension, and the root cause analysis result of the target indicator is obtained based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator. Thus, the target indicator can be subjected to single-dimensional root cause analysis to obtain the first dimension item, and the target indicator can also be subjected to multi-dimensional root cause analysis to obtain the combination of the second dimension items, and the root cause analysis result of the target indicator is obtained by comprehensively considering the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result, thereby improving the comprehensiveness of the root cause analysis.

[0207] In order to implement the above embodiments, the present disclosure further proposes a computer-readable storage medium on which computer program instructions are stored. When the program instructions are executed by a processor, the steps of the root cause analysis method provided by the present disclosure are implemented.

[0208] Alternatively, the computer readable storage medium may be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, or the like.

[0209] In order to implement the above embodiments, the present disclosure further provides a computer program product, including a computer program, characterized in that when the computer program is executed by a processor of an electronic device, the root cause analysis method as described above is implemented.

[0210] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0211] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A root cause analysis method, characterized in that: include: Acquire a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator; Based on the data of the plurality of candidate dimension items, a first dimension item is screened out from the plurality of candidate dimension items as a single-dimensional root cause analysis result of the target indicator; Based on the data of the plurality of candidate dimension items, a combination of second dimension items is screened out from the combination of the plurality of candidate dimension items as a multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension; Based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator, the root cause analysis result of the target indicator is obtained.

2. The method according to claim 1, characterized in that The step of selecting a first dimension item from the plurality of candidate dimension items based on the data of the plurality of candidate dimension items includes: Determining a sixth dimension to be deleted from the plurality of candidate dimensions based on at least one of the number of candidate dimension items under the candidate dimension, the name of the candidate dimension, and a user's preference for the candidate dimension; Deleting the sixth dimension from the plurality of candidate dimensions, and determining the plurality of candidate dimensions remaining after the deletion as the first dimension; Based on the data of multiple candidate dimension items under the first dimension, the first dimension item is filtered out from the multiple candidate dimension items under the first dimension.

3. The method according to claim 2, characterized in that The determining a sixth dimension to be deleted from the plurality of candidate dimensions based on at least one of the number of candidate dimension items under the candidate dimensions, the name of the candidate dimension, and the user's preference for the candidate dimension comprises: If the number of candidate dimension items under the candidate dimension is greater than or equal to the first set threshold, or the number of candidate dimension items under the candidate dimension is less than or equal to the second set threshold, the candidate dimension is determined as the sixth dimension; and / or, If the name of the candidate dimension is a set name, determining the candidate dimension as the sixth dimension; and / or, If the user's preference for the candidate dimension is less than or equal to a third set threshold, the candidate dimension is determined as the sixth dimension.

4. The method according to claim 1, characterized in that: The step of selecting a first dimension item from the plurality of candidate dimension items based on the data of the plurality of candidate dimension items includes: Based on the data of the candidate dimension items, obtaining the explanatory power and surprise degree of the candidate dimension items; Sort the plurality of candidate dimension items from large to small according to the absolute value of the explanatory power, add the top N candidate dimension items to the target set, and / or, add the candidate dimension items whose absolute value of the explanatory power is greater than a fourth set threshold to the target set, where N is a positive integer; The multiple candidate dimension items in the target set are sorted from large to small according to the absolute value of the surprise degree, and the top M candidate dimension items are determined as the first dimension items, where M is a positive integer.

5. The method according to claim 1, characterized in that: The step of selecting a first dimension item from the plurality of candidate dimension items based on the data of the plurality of candidate dimension items includes: Based on the actual value and reference value of the candidate dimension item at the current moment, and the actual value and reference value of the target indicator at the current moment, the contribution of the candidate dimension item is obtained; The plurality of candidate dimension items are sorted from large to small according to the absolute values ​​of the contribution degrees, and the top M candidate dimension items are determined as the first dimension items, where M is a positive integer.

6. The method according to claim 1, characterized in that The step of screening out a combination of second dimension items from a combination of the plurality of candidate dimension items based on the data of the plurality of candidate dimension items includes: Based on the data of the plurality of candidate dimension items, a combination of the third dimension items of the i-th round is screened out from the combination of the plurality of candidate dimension items, wherein i is a positive integer and the initial value of i is 1; Determining whether the number of combinations of the third dimension items in the i-th round is less than or equal to a fifth set threshold; If the number of combinations of the third dimension items in the i-th round is less than or equal to the fifth set threshold, the combinations of the third dimension items in the i-th round are used as combinations of fourth dimension items, and based on the combinations of the fourth dimension items, combinations of the second dimension items are obtained; If the number of combinations of the third dimension items in the i-th round is greater than the fifth set threshold, continue to execute the screening process of the i+1th round until the number of combinations of the third dimension items in the target round is less than or equal to the fifth set threshold, take the combinations of the third dimension items in the target round as the combinations of the fourth dimension items, and based on the combinations of the fourth dimension items, obtain the combinations of the second dimension items.

7. The method according to claim 6, characterized in that The step of screening out a combination of third dimension items of the i-th round from a combination of the plurality of candidate dimension items based on the data of the plurality of candidate dimension items comprises: If i=1, based on the data of the first dimension items under any two candidate dimensions, a combination of the third dimension items of the first round is screened out from the combination of the first dimension items under the any two candidate dimensions.

8. The method according to claim 6, characterized in that The step of screening out a combination of third dimension items of the i-th round from a combination of the plurality of candidate dimension items based on the data of the plurality of candidate dimension items includes: If i ≥ 2, the combination of the third dimension items of the i-1th round is split to obtain multiple third dimension items of the i-1th round; Taking the multiple third dimension items of the (i-1)th round as fifth dimension items; Based on the data of the fifth dimension items under any i+1 candidate dimensions, a combination of the third dimension items of the i-th round is screened out from the combination of the fifth dimension items under any i+1 candidate dimensions.

9. The method according to claim 6, characterized in that The obtaining a combination of the second dimensional items based on the combination of the fourth dimensional items includes: Splitting the combination of the fourth dimension items to obtain a plurality of fourth dimension items; Based on the data of the plurality of fourth dimension items, a combination of the second dimension items is screened out from the combinations of the plurality of fourth dimension items.

10. The method according to any one of claims 1 to 9, characterized in that The data of the candidate dimension item includes a reference value of the candidate dimension item at the current moment, and the method further includes: Obtaining actual values ​​of the candidate dimension items at multiple historical moments; The average value of the actual values ​​of the candidate dimension item at multiple historical moments is used as the reference value of the candidate dimension item at the current moment; or The actual values ​​of the candidate dimension items at multiple historical moments are weighted and summed to obtain the reference value of the candidate dimension items at the current moment.

11. The method according to any one of claims 1 to 9, characterized in that The root cause analysis result of the target indicator is obtained based on the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator, including: Normalizing the contribution of the single-dimensional root cause analysis results and the multi-dimensional root cause analysis results of the target indicator; The single-dimensional root cause analysis results and multi-dimensional root cause analysis results of the target indicator are sorted from large to small according to the absolute value of the contribution after normalization, and the first Q results are used as the root cause analysis results of the target indicator, where Q is a positive integer.

12. The method according to claim 11, characterized in that The contribution of the multidimensional root cause analysis results of the target indicator is normalized, including: Splitting the combination of the second dimension items to obtain a plurality of second dimension items; Based on the number of candidate dimensions to which the plurality of second dimension items belong, updating the contribution of the multidimensional root cause analysis result of the target indicator; The updated contribution of the multidimensional root cause analysis result of the target indicator is normalized.

13. A root cause analysis device, characterized in that: include: A first acquisition module is configured to execute acquisition of a target indicator of a product and candidate dimension items of the target indicator under multiple candidate dimensions, wherein all candidate dimension items of the target indicator under any candidate dimension are used to aggregate into the target indicator; A first screening module is configured to perform data based on the plurality of candidate dimension items, and screen out a first dimension item from the plurality of candidate dimension items as a single-dimensional root cause analysis result of the target indicator; A second screening module is configured to perform, based on the data of the plurality of candidate dimension items, screening out a combination of second dimension items from the combination of the plurality of candidate dimension items as a multi-dimensional root cause analysis result of the target indicator, wherein each second dimension item in the combination of the second dimension items belongs to a different candidate dimension; The second acquisition module is configured to execute the single-dimensional root cause analysis result and the multi-dimensional root cause analysis result of the target indicator to obtain the root cause analysis result of the target indicator.

14. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to: Implement the steps of the method according to any one of claims 1 to 12.

15. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the program instructions are executed by a processor, the steps of the method described in any one of claims 1 to 12 are implemented.

Images