Data fluctuation root cause determination method and device, electronic equipment and storage medium

By calculating and arranging the combination of fundamental factors to be selected in the business report, the combination of fundamental factors targeted for data fluctuations during adjacent observation weeks is determined, and the problems of low manual positioning efficiency and poor accuracy are solved, and accurate analysis of data fluctuations and strategy adjustment are achieved. .

CN118917692AInactive Publication Date: 2024-11-08中国联合网络通信有限公司广东省分公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411056085.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-08
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

When data fluctuations are abnormal in the existing technology, the artificial positioning fundamental factors are inefficient and easily missed, resulting in inaccurate adjustment strategies.

Method used

By obtaining the data of two adjacent observation periods, the root cause evaluation index value of each selected fundamental factor combination is calculated, and the combination of fundamental factor to be selected is arranged according to the factor dimension type to determine the target fundamental factor combination that causes data fluctuations.

Benefits of technology

It improves the determination efficiency and accuracy of the combination of fundamental factors of data fluctuations to ensure the accuracy of strategy adjustments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118917692A_ABST
    Figure CN118917692A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses a data fluctuation root cause determination method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring first observation data of a first observation period and second observation data of a second observation period; the first observation data comprises a first key index value corresponding to each to-be-selected root factor combination, and the second observation data comprises a second key index value corresponding to each to-be-selected root factor combination; determining a root cause evaluation index value of each to-be-selected root factor combination based on the first key index value and the second key index value corresponding to the same to-be-selected root factor combination; based on the root cause evaluation index value and the factor dimension type of each to-be-selected root factor combination, arranging the to-be-selected root factor combinations to determine a to-be-selected root factor combination sequence; therefore, based on the to-be-selected root factor combination sequence of each factor dimension type, the target root factor combination causing data fluctuation between the two adjacent observation periods is accurately and conveniently determined.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of data analysis technology, and in particular, to a method, device, electronic device, and storage medium for determining a root cause of data fluctuation. Background Art

[0002] As the trend of digital transformation of enterprises gradually develops, data analysis technology occupies a high position in enterprise business decision-making. Usually, enterprises need to monitor a large number of key indicators in business reports, analyze anomalies or explore the causes of fluctuations in order to adjust marketing strategies in a timely manner.

[0003] When existing companies discover that key indicators in operating statements have abnormal fluctuations, they usually need to manually locate the root factors that cause data fluctuations. However, if the root factors that cause data fluctuations are located manually, the efficiency is low, and because the data in the operating statements have multiple dimensions, manual root cause location is prone to miss certain factor combinations, resulting in inaccurate adjustment strategies based on inaccurate root factors, and unable to effectively adjust data fluctuations. Therefore, how to accurately and conveniently determine the root factors that cause data fluctuations is a problem that needs to be solved urgently. Summary of the invention

[0004] Embodiments of the present invention provide a method, device, electronic device and storage medium for determining the root cause of data fluctuations, so as to accurately and conveniently determine the combination of fundamental factors that cause data fluctuations between two adjacent observation periods, thereby improving the efficiency and accuracy of determining the combination of fundamental factors.

[0005] In a first aspect, an embodiment of the present invention provides a method for determining a root cause of data fluctuation, comprising:

[0006] Acquire first observation data corresponding to a first observation period and second observation data corresponding to a second observation period in two adjacent observation periods, wherein the first observation data includes a first key indicator value corresponding to each candidate fundamental factor combination, and the second observation data includes a second key indicator value corresponding to each candidate fundamental factor combination;

[0007] Determine the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination;

[0008] Based on the root cause evaluation index value and the factor dimension type corresponding to each candidate root factor combination, the candidate root factor combinations corresponding to each factor dimension type are arranged to determine a sequence of candidate root factor combinations corresponding to each factor dimension type;

[0009] Based on the sequence of candidate fundamental factor combinations corresponding to each factor dimension type, a target fundamental factor combination causing data fluctuations between the two adjacent observation periods is determined.

[0010] In a second aspect, an embodiment of the present invention further provides a device for determining a root cause of data fluctuation, the device comprising:

[0011] An observation data acquisition module, used to acquire first observation data corresponding to a first observation period and second observation data corresponding to a second observation period in two adjacent observation periods, wherein the first observation data includes a first key indicator value corresponding to each candidate root factor combination, and the second observation data includes a second key indicator value corresponding to each candidate root factor combination;

[0012] A root cause evaluation index value determination module, used to determine the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination;

[0013] A candidate root factor combination sequence determination module is used to arrange the candidate root factor combinations corresponding to each factor dimension type based on the root cause evaluation index value and the factor dimension type corresponding to each candidate root factor combination, and determine the candidate root factor combination sequence corresponding to each factor dimension type;

[0014] The target fundamental factor combination determination module is used to determine the target fundamental factor combination that causes data fluctuations between the two adjacent observation periods based on the sequence of candidate fundamental factor combinations corresponding to each factor dimension type.

[0015] In a third aspect, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0016] one or more processors;

[0017] A memory for storing one or more programs;

[0018] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining the root cause of data fluctuation as provided in any embodiment of the present invention.

[0019] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for determining a root cause of data fluctuations as provided in any embodiment of the present invention.

[0020] In a fifth aspect, an embodiment of the present invention provides a computer program product, including a computer program, which, when executed by a processor, implements a method for determining a root cause of data fluctuations as provided in any embodiment of the present invention.

[0021] The technical solution of the embodiment of the present invention is to obtain first observation data corresponding to a first observation period and second observation data corresponding to a second observation period in two adjacent observation periods, wherein the first observation data includes a first key indicator value corresponding to each candidate root factor combination, and the second observation data includes a second key indicator value corresponding to each candidate root factor combination; based on the first key indicator value and the second key indicator value corresponding to the same candidate root factor combination, determine the root cause evaluation indicator value corresponding to each candidate root factor combination, so that the data fluctuation between the first key indicator value and the second key indicator value of each candidate root factor combination in two adjacent observation periods is accurately quantified according to a unified standard, thereby improving the accuracy of determining the root cause evaluation indicator value; based on the root cause evaluation indicator value and the factor dimension type corresponding to each candidate root factor combination, arrange the candidate root factor combinations corresponding to each factor dimension type, and determine the candidate root factor combination sequence corresponding to each factor dimension type; thereby, based on the candidate root factor combination sequence corresponding to each factor dimension type, accurately and conveniently determine the target root factor combination that causes the data fluctuation between the two adjacent observation periods, thereby improving the efficiency and accuracy of determining the target root factor combination.

[0022] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 is a flow chart of a method for determining a root cause of data fluctuation provided in Embodiment 1 of the present invention;

[0025] Figure 2 is a flow chart of a method for determining a root cause of data fluctuation provided by Embodiment 2 of the present invention;

[0026] Figure 3 is a line graph of a comprehensive score in a two-dimensional type involved in the second embodiment of the present invention;

[0027] Figure 4 is a two-dimensional type of contribution line graph involved in the second embodiment of the present invention;

[0028] Figure 5 is a line graph of a comprehensive score in a three-dimensional type involved in the second embodiment of the present invention;

[0029] Figure 6 is a line graph of contribution in a three-dimensional type involved in the second embodiment of the present invention;

[0030] Figure 7 1 is a schematic diagram of the structure of a device for determining a root cause of data fluctuation provided by Embodiment 3 of the present invention;

[0031] Figure 8 It is a schematic diagram of the structure of an electronic device for implementing the method for determining the root cause of data fluctuation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0033] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention 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 invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0034] Embodiment 1

[0035] Figure 1 A flowchart of a method for determining the root cause of data fluctuation is provided for the first embodiment of the present invention. This embodiment is applicable to the case of determining the target root factor combination that causes data fluctuation between two adjacent observation periods. The method can be executed by a data fluctuation root cause determination device, which can be implemented in the form of hardware and / or software, and can be configured in an electronic device. Figure 1 As shown, the method includes:

[0036] S110, acquiring first observation data corresponding to a first observation period and second observation data corresponding to a second observation period in two adjacent observation periods.

[0037] The first observation data includes the first key indicator value corresponding to each candidate fundamental factor combination, and the second observation data includes the second key indicator value corresponding to each candidate fundamental factor combination. The candidate fundamental factor combination may refer to all fundamental factor combinations that affect the stability of the key indicator value. The candidate fundamental factor combination may be composed of any fundamental factor under at least one dimension in the candidate dimension combination. For example, the candidate fundamental factor combination may be composed of "channel 1" or "channel 2" under the "channel" dimension in the candidate dimension combination, and the factor dimension type of the candidate fundamental factor combination (channel 1 and channel 2) is a one-dimensional type. The candidate fundamental factor combination may also be composed of "channel 1" or "channel 2" under the "channel" dimension, and "product 1" or "product 2" under the "product" dimension, and the factor dimension type of the candidate fundamental factor combination (channel 1 & product 1, channel 1 & product 2, channel 2 & product 1 and channel 2 & product 2) is a two-dimensional type. Key indicators may refer to key performance indicators (KPIs). Key indicators may be pre-set based on user business needs. Two adjacent observation periods may be, but are not limited to, a previous observation period and a current observation period. The first observation period may be the previous observation period. The second observation period may be the current observation period.

[0038] Specifically, various observation data in the current observation period can be collected regularly, and the second key indicator value corresponding to each candidate fundamental factor combination can be calculated, and the first key indicator value corresponding to each candidate fundamental factor combination in the previous observation period can be obtained.

[0039] It should be noted that the first observation data may also include a first basic data set. The second observation data may also include a second basic data set. The first basic data set may include a data set for calculating a first key indicator value and a combination of dimensions to be selected. The second basic data set may include a data set for calculating a second key indicator value and a combination of dimensions to be selected. The role of the basic data sets is to be arbitrarily combined to obtain key indicator values ​​corresponding to different fundamental factor combinations.

[0040] Among them, the first basic data set and the first key indicator value of the previous observation period, as well as the second basic data set and the second key indicator value of the current observation period are obtained. Key indicators can be divided into quantitative indicators and rate indicators. The advantage of such a setting is that the method for determining the root cause of data fluctuations proposed in the present invention can be applied to both quantitative indicators and rate indicators, thereby increasing universality. The quantitative indicator may refer to an additive indicator, that is, the overall indicator is equal to the sum of the sub-item indicators. For example, the quantitative indicator may be but is not limited to production capacity and visit volume. The rate indicator may be an indicator derived from the quantitative indicator. The rate indicator is not additive, that is, the sub-item indicators cannot be directly added. For example, the rate indicator may be but is not limited to click-through rate and unsubscription rate. The rate indicator may be determined using the quantitative indicator. For example, the rate indicator may be defined as P=X / M, where X and M are both quantitative indicators. Table 1 gives an example of observation data.

[0041] Table 1 Example of observation data

[0042]

[0043] Referring to Table 1, the key indicator is defined as a rate value indicator, namely, the product new development cancellation rate. Product new development cancellation rate = new development cancellation scale / new development scale. Dimensions may include "product", "channel" and "city". The fundamental factors under the dimension may include "product 1", "channel 1", "city 1" and "city 2". The dimension combination to be selected may include one-dimensional, two-dimensional and three-dimensional dimension combinations. "Product", "channel" or "city" is a one-dimensional dimension combination. "Product & channel", "product & city" or "channel & city" is a two-dimensional dimension combination. "Product & channel & city" is a three-dimensional dimension combination. "City 1" and "City 2" are the fundamental factors under the "city" dimension combination. "Product 1 & channel 1" is the fundamental factor combination under the "product & channel" dimension combination.

[0044] Specifically, if the observed data involves n dimensions, the first key indicator value and the second key indicator value corresponding to each candidate fundamental factor combination can be determined based on the first basic data set and the second basic data set.

[0045] It should be noted that the data of the two observation cycles can also be replaced by expected completion data and actual completion data to evaluate the completion of KPIs, further improving the versatility of the method for determining the root causes of data fluctuations.

[0046] S120: Determine the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination.

[0047] The root cause evaluation index value may refer to the influence of the root factor combination on the data fluctuation between two adjacent observation periods determined based on a unified standard. The selected root factor combination may be H root factor combinations decomposed from the first basic data set based on the Mutually Exclusive Collectively Exhaustive (MECE) principle.

[0048] On the basis of the above technical solution, the root cause evaluation index value includes: at least one of contribution, deviation and comprehensive score; "determining the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination" includes at least one of the following:

[0049] Determine the first key indicator difference based on the first key indicator value and the second key indicator value corresponding to the same candidate fundamental factor combination; determine the second key indicator difference based on the total value of the first key indicator and the total value of the second key indicator corresponding to the candidate dimension combination to which the same candidate fundamental factor combination belongs; determine the contribution degree corresponding to the same candidate fundamental factor combination by dividing the first key indicator difference and the second key indicator difference.

[0050] Among them, the contribution EP can be used to reflect the contribution of the fluctuation of the candidate fundamental factor combination to the fluctuation of the key indicator. The contribution EP satisfies Σ i∈H EP i =1. A larger value of the contribution EP indicates that the contribution of the candidate fundamental factor combination to the key indicator fluctuation is greater. The candidate dimension combination may include at least one candidate fundamental factor combination. For example, "prefecture-level city" may include "prefecture-level city 1" and "prefecture-level city 2".

[0051] Specifically, the second key indicator value of the i-th candidate fundamental factor combination in the current observation period is The first key indicator value of the i-th candidate fundamental factor combination in the previous observation period is The contribution EP formula is:

[0052]

[0053] in, Indicates the second key indicator value, Indicates the first key indicator value, Indicates the difference of the first key indicator, X 1 Indicates the total value of the second key indicator, X 0 Indicates the total value of the first key indicator, X 1 -X 0 Indicates the second key indicator difference.

[0054] Determine the first key indicator ratio based on the total value of the first key indicator and the first key indicator value corresponding to the same candidate fundamental factor combination; determine the second key indicator ratio based on the total value of the second key indicator and the second key indicator value corresponding to the same candidate fundamental factor combination; determine the deviation corresponding to the same candidate fundamental factor combination based on the difference between the second key indicator ratio and the first key indicator ratio and the sum of the second key indicator ratio and the first key indicator ratio.

[0055] Among them, the deviation S can be used to reflect the structural fluctuation degree of the value of the candidate fundamental factor combination. The closer the deviation S is to 0, the smaller the change of the value of the candidate fundamental factor combination in two adjacent observation periods. The deviation S equals 0, which indicates that the value of the candidate fundamental factor combination has no change in two adjacent observation periods.

[0056] Specifically, using the above example, the formula for the deviation S is:

[0057]

[0058] in, Represents the first key indicator ratio, Indicates the ratio of the second key indicator.

[0059] The contribution and deviation corresponding to the same candidate fundamental factor combination are multiplied to determine the comprehensive score corresponding to the same candidate fundamental factor combination.

[0060] Among them, the comprehensive score SCORE = EP*abs(S), where abs() represents the absolute value. When the contribution EP and the deviation S are both larger, it means that the candidate fundamental factor combination is more likely to be the target fundamental factor combination that causes data fluctuations between two adjacent observation periods.

[0061] In S120, specifically, taking the rate value index as an example, the above example is used, and the rate value index value is P = X / M, where X and M are both quantitative indexes. The key index value corresponding to the i-th candidate fundamental factor combination is P i =X i / M i ,and Defining Contribution EP i =A i +B i . Determine A i and B i The specific formula is:

[0062]

[0063]

[0064] in, It represents the second rate index value corresponding to the i-th candidate fundamental factor combination, represents the first rate index value corresponding to the i-th candidate fundamental factor combination, P 1 P represents the total value of the second rate value index corresponding to the candidate dimension combination to which the i-th candidate fundamental factor combination belongs, 0 It represents the total value of the first rate value index corresponding to the candidate dimension combination to which the i-th candidate fundamental factor combination belongs. represents the first quantitative index value corresponding to the i-th candidate fundamental factor combination, represents the second quantitative index value corresponding to the i-th candidate fundamental factor combination, M 0 M represents the total value of the first quantitative index corresponding to the candidate dimension combination to which the i-th candidate fundamental factor combination belongs, 1 It represents the total value of the second quantitative index corresponding to the candidate dimension combination to which the i-th candidate fundamental factor combination belongs.

[0065] The formula for deviation S is:

[0066]

[0067] Comprehensive score SCORE = EP*abs(S), where abs() represents the absolute value.

[0068] S130. Based on the root cause evaluation index value and the factor dimension type corresponding to each candidate root factor combination, arrange the candidate root factor combinations corresponding to each factor dimension type to determine a sequence of candidate root factor combinations corresponding to each factor dimension type.

[0069] The factor dimension type may refer to the dimension type of the candidate dimension combination to which the candidate root factor combination belongs. For example, the factor dimension type may be, but is not limited to, a one-dimensional type, a two-dimensional type, or a multi-dimensional type (including three-dimensional and above types). Each element in the candidate root factor combination sequence may include the candidate root factor combination and the root cause evaluation index value corresponding to the candidate root factor combination.

[0070] Specifically, all candidate root factor combinations are classified according to factor dimension types, and candidate root factor combinations corresponding to each factor dimension type and root cause evaluation index values ​​corresponding to the candidate root factor combinations are determined. The candidate root factor combinations are arranged in descending order according to the root cause evaluation index values ​​corresponding to the candidate root factor combinations contained in each factor dimension type, and a sequence of candidate root factor combinations corresponding to each factor dimension type is obtained.

[0071] S140. Based on the sequence of candidate fundamental factor combinations corresponding to each factor dimension type, determine the target fundamental factor combination that causes data fluctuations between two adjacent observation periods.

[0072] The target fundamental factor combination may refer to a candidate fundamental factor combination with the highest probability of causing data fluctuations between two adjacent observation periods.

[0073] Specifically, the first candidate root factor combination sequence in the candidate root factor combination sequence corresponding to each factor dimension type is determined, and all first candidate root factor combination sequences are determined as target root factor combinations that cause data fluctuations between two adjacent observation periods.

[0074] The technical solution of the embodiment of the present invention is to obtain first observation data corresponding to the first observation period and second observation data corresponding to the second observation period in two adjacent observation periods, wherein the first observation data includes the first key indicator value corresponding to each candidate root factor combination, and the second observation data includes the second key indicator value corresponding to each candidate root factor combination; based on the first key indicator value and the second key indicator value corresponding to the same candidate root factor combination, determine the root cause evaluation indicator value corresponding to each candidate root factor combination, so that the data fluctuation between the first key indicator value and the second key indicator value of each candidate root factor combination in two adjacent observation periods is accurately quantified according to a unified standard, thereby improving the accuracy of determining the root cause evaluation indicator value; based on the root cause evaluation indicator value and the factor dimension type corresponding to each candidate root factor combination, arrange the candidate root factor combinations corresponding to each factor dimension type, and determine the candidate root factor combination sequence corresponding to each factor dimension type; thereby, based on the candidate root factor combination sequence corresponding to each factor dimension type, accurately and conveniently determine the target root factor combination that causes data fluctuation between two adjacent observation periods, thereby improving the efficiency and accuracy of determining the target root factor combination.

[0075] Embodiment 2

[0076] Figure 2 This is a flowchart of a method for determining the root cause of data fluctuations provided in the second embodiment of the present invention. Based on the above embodiment, this embodiment describes in detail the process of determining the target root factor combination that causes data fluctuations between two adjacent observation periods. The explanations of the terms that are the same or corresponding to the above embodiments are not repeated here. Figure 2 As shown, the method includes:

[0077] S210, obtaining first observation data corresponding to a first observation period and second observation data corresponding to a second observation period in two adjacent observation periods.

[0078] The first observation data includes a first key indicator value corresponding to each candidate fundamental factor combination, and the second observation data includes a second key indicator value corresponding to each candidate fundamental factor combination.

[0079] S220: Determine the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination.

[0080] It should be noted that after determining the comprehensive score in the root cause evaluation index value corresponding to each candidate root factor combination, it is also possible to traverse all candidate dimension combinations, calculate the variance SCORE_VAR of the comprehensive score SCORE of each candidate root factor combination under each candidate dimension, sort the variance SCORE_VAR from high to low, and the candidate dimension combination with the largest SCORE_VAR is judged as the dimension combination that most affects the KPI fluctuation.

[0081] Table 2 gives an example of the variance ranking of each dimension under a candidate dimension combination.

[0082] Table 2 Examples of variance rankings for each dimension

[0083] Dimension combination SCORE_VAR channel 0.0006 product 0.0001 City 0.0001

[0084] As shown in Table 2, “channel” is the most important dimension combination in the one-dimensional type that affects the fluctuation of key indicators.

[0085] S230: Based on the root cause evaluation index value and the factor dimension type corresponding to each candidate root factor combination, arrange the candidate root factor combinations corresponding to each factor dimension type, and determine the candidate root factor combination sequence corresponding to each factor dimension type.

[0086] S240. If the dimension types of each factor are all one-dimensional, then based on each candidate root factor combination in the sequence of candidate root factor combinations corresponding to the one-dimensional type and the root cause evaluation index value of each candidate root factor combination, determine the one-dimensional root factor combination, and use the one-dimensional root factor combination as the target root factor combination that causes data fluctuations between two adjacent observation periods.

[0087] Specifically, for the preset thresholds of contribution EP, deviation S and comprehensive score SCORE in the root cause evaluation index, a one-dimensional root factor combination greater than the preset threshold is determined from each candidate root factor combination in the candidate root factor combination sequence corresponding to the one-dimensional type, and the one-dimensional root factor combination is used as the target root factor combination that causes data fluctuations between two adjacent observation periods. Table 3 gives an example of a one-dimensional root factor combination, as follows:

[0088] Table 3 An example of a one-dimensional combination of fundamental factors

[0089]

[0090] As shown in Table 3, the one-dimensional fundamental factor combination can display the target fundamental factor combination, the corresponding dimensions of the target fundamental factor combination, and the root cause evaluation indicators corresponding to the target fundamental factor combination, including contribution EP, deviation S, comprehensive score SCORE, etc.

[0091] On the basis of the above technical scheme, "determining a one-dimensional root factor combination based on each candidate root factor combination in the sequence of candidate root factor combinations corresponding to the one-dimensional type and the root cause evaluation index value of each candidate root factor combination" may include: determining the maximum absolute value of the difference in root cause evaluation index values ​​between adjacent candidate root factor combinations in the sequence of candidate root factor combinations corresponding to the one-dimensional type; determining the one-dimensional root factor combination based on the previous candidate root factor combination in the adjacent candidate root factor combinations corresponding to the maximum absolute value of the difference and each candidate factor combination before the candidate root factor combination in the sequence of candidate root factor combinations.

[0092] The candidate fundamental factor combination may refer to the fundamental factor combination located at the previous position among the adjacent candidate fundamental factor combinations corresponding to the maximum absolute value of the difference.

[0093] Specifically, in the sequence of candidate root factor combinations corresponding to the one-dimensional type, the root cause evaluation index value of the next candidate root factor combination is subtracted from the root cause evaluation index value of the previous candidate root factor combination to determine the difference between the root cause evaluation index values ​​of each adjacent candidate root factor combination, and the absolute value of the difference is taken to obtain the absolute value of the difference, so as to determine the maximum absolute value of the difference and the adjacent candidate root factor combination corresponding to the maximum absolute value of the difference from each absolute value of the difference. The candidate root factor combination located in the previous position in the adjacent candidate root factor combination and each candidate factor combination before the candidate root factor combination in the sequence of candidate root factor combinations are determined as a one-dimensional root factor combination, so that there is no need to manually set a threshold, and the whole process is automated without the need for subjective human intervention, further improving the accuracy of determining the one-dimensional root factor group.

[0094] On the basis of the above technical scheme, "determining the maximum absolute value of the difference in root cause evaluation index values ​​between adjacent candidate root factor combinations in the sequence of candidate root factor combinations corresponding to the one-dimensional type" may include: determining a line graph based on the root cause evaluation index values ​​corresponding to a preset number of candidate root factor combinations in the sequence of candidate root factor combinations corresponding to the one-dimensional type; determining the maximum absolute value of the difference in root cause evaluation index values ​​between adjacent candidate root factor combinations in the sequence of candidate root factor combinations corresponding to the one-dimensional type based on the candidate root factor combination with the largest corner in the line graph.

[0095] The pre-set number may refer to the preset number of horizontal coordinates Q of the line graph. Q may be any integer greater than 1. The root cause evaluation index T may be selected as any one or more of the contribution EP, the deviation S, and the comprehensive score SCORE. If multiple are selected, the root causes may be determined by the elbow method respectively, and then the root causes may be combined.

[0096] Specifically, according to the elbow method, the inflection point N in the line graph is selected as the threshold, and a one-dimensional root factor combination is determined based on the inflection point N and the candidate root factor combination before the inflection point N. A line graph is drawn by selecting the comprehensive score in the root cause evaluation index, with the X-axis being the value 1 to Q and the Y-axis being the root cause evaluation index T, such as the comprehensive score. A first-order difference operation is performed on the candidate root factor combination sequence Y to obtain a new candidate root factor combination sequence Z. For example, assuming that the candidate root factor combination sequence Y is Y1, Y2, ..., YQ, then Z1 = Y2-Y1, Z2 = Y3-Y2, ..., Z(Q-1) = YQ-Y(Q-1). The maximum absolute value of the candidate root factor combination sequence Z is taken, and the position of the maximum absolute value is the threshold. For example, if the maximum absolute value is ZN, a one-dimensional root factor combination is determined based on 1 to N sequences in the candidate root factor combination sequence. For example, if Q = 10, the root cause evaluation indicators include the comprehensive score SCORE and the contribution EP, then according to the elbow method, the threshold of the comprehensive score SCORE is selected as N = 3, and the threshold of the contribution EP is selected as N = 1. Thus, the root factor combination is screened by the elbow method, and the whole process is automated without the need for subjective human intervention, further improving the accuracy of determining the one-dimensional root factor group.

[0097] S250. If there are two-dimensional types or multi-dimensional types in each factor dimension type, then based on the low-dimensional fundamental factor combination corresponding to the low-dimensional type and other fundamental factor combinations, determine the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type, and based on the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type, determine the high-dimensional fundamental factor combination; other fundamental factor combinations are fundamental factor combinations under other candidate dimension combinations except the candidate dimension combinations to which the low-dimensional fundamental factor combinations belong.

[0098] Among them, the factor dimension type can be but not limited to one-dimensional type, two-dimensional type, multi-dimensional type (including three-dimensional and above types). High dimension and low dimension are relative. Two dimensions are high dimensions relative to one dimension. Two dimensions are low dimensions relative to three dimensions.

[0099] Specifically, taking the three-dimensional type as an example, based on the above method, such as the line graph, the one-dimensional fundamental factor combination is determined as shown in Table 4, as follows:

[0100] Table 4 Another example of one-dimensional fundamental factor combination

[0101]

[0102] As shown in Table 4, the one-dimensional root factor combination is channel 1, product 1, and channel 2, thereby narrowing the root cause location range of abnormal fluctuations. After determining the one-dimensional root factor combination, the dimensions of the one-dimensional root factor combination can be deduplicated to obtain a one-dimensional root cause dimension set {W1} (equivalent to the dimension combination to be selected), and the factors of the one-dimensional root factor combination can be deduplicated to obtain a one-dimensional root cause factor set {V1}.

[0103] Specifically, all elements (equivalent to dimensions) Wi of the one-dimensional root cause dimension set {W1} are traversed. Each time, Wi is taken as the first dimension, and the element values ​​of Wi that are not in the one-dimensional root cause factor set {V1} are eliminated (equivalent to the fundamental factors under the dimension). For each Wi, the remaining dimensions except Wi are traversed as the second dimension. According to the basic data set, the key indicator values ​​corresponding to each combination of fundamental factors under the first dimension and the second dimension are obtained by processing, and then the root cause evaluation indicator values ​​under each combination of fundamental factors are calculated. It is also possible to select the root cause evaluation indicator values ​​corresponding to each combination of fundamental factors under the first dimension and the second dimension from the root cause evaluation indicator values ​​corresponding to each combination of fundamental factors determined in advance, that is, the root cause evaluation indicator values ​​corresponding to the combination of fundamental factors of the high-dimensional type determined based on the low-dimensional fundamental factor combination of the low-dimensional type and other fundamental factor combinations. The newly determined root factor combination is deduplicated to obtain a sequence of combinations of fundamental factors of the high-dimensional type corresponding to the combination of fundamental factors to be selected. For example, the fundamental factor combination {the fundamental factor combination of the first dimension = (Wi, Vj), the fundamental factor combination of the second dimension = (Wm, Vn)} and {the fundamental factor combination of the first dimension = (Wm, Vn), the fundamental factor combination of the second dimension = (Wi, Vj)} belong to the same candidate fundamental factor combination, and only one of them needs to be retained.

[0104] Continuing with the above example, if Q=10, the root cause evaluation indicators include the comprehensive score SCORE and the contribution EP, then draw a line graph and select the inflection point N as the threshold according to the elbow method. Figure 3 A line chart of the comprehensive score in a two-dimensional type is given. Figure 4 A two-dimensional contribution line graph is given. Figure 3 and Figure 4 , the threshold of the comprehensive score SCORE is selected as N = 2, and the threshold of the contribution EP is selected as N = 1. Table 5 gives an example of a two-dimensional fundamental factor combination, as follows:

[0105] Table 5 Examples of two-dimensional fundamental factor combinations

[0106]

[0107] See Table 5, the two-dimensional fundamental factor combinations are channel 2 & product 2, channel 1 & product 1. After determining the two-dimensional fundamental factor combination, the first dimension of the two-dimensional fundamental factor combination can be deduplicated to obtain the first dimension set {W1}, and the first factor of the two-dimensional fundamental factor combination can be deduplicated to obtain the first factor set {V1}. Deduplication of the second dimension of the two-dimensional fundamental factor combination is performed, and the dimensions in the first dimension set {W1} are eliminated to obtain the second dimension set {W2}. Deduplication of the second factor of the two-dimensional fundamental factor combination is performed, and the factors in the first factor set {V1} are eliminated to obtain the second factor set {V2}. This ensures that {W1} and {W2} have no intersection, and {V1} and {V2} have no intersection.

[0108] Specifically, all elements Wi of the first dimension {W1} are traversed, and each traversal takes Wi as the first dimension, and the element values ​​of Wi that are not in the first factor set {V1} are eliminated. For each Wi, all elements Wj of the second dimension {W2} are traversed, and each traversal takes {Wi, Wj} as the first and second dimensions, and the element values ​​of Wj that are not in the second factor set {V2} are eliminated. For each {Wi, Wj}, the remaining dimensions except Wi and Wj are traversed as the third dimension. The root cause evaluation indicators of each combination of fundamental factors under the first, second, and third dimensions are determined, and the filtered combinations of fundamental factors are deduplicated. For example, the fundamental factor combination {the fundamental factor combination of the first dimension = (Wi, Vj), the fundamental factor combination of the second dimension = (Wm, Vn), the fundamental factor combination of the third dimension = (Wk, Vr)} and {the fundamental factor combination of the first dimension = (Wm, Vn), the fundamental factor combination of the second dimension = (Wk, Vr), the fundamental factor combination of the third dimension = (Wi, Vj)} belong to the same candidate fundamental factor combination, and you only need to keep one.

[0109] Continuing with the above example, if Q=10, the root cause evaluation indicators include the comprehensive score SCORE and the contribution EP, then draw a line graph and select the inflection point N as the threshold according to the elbow method. Figure 5 A line chart of the comprehensive score in a three-dimensional type is given. Figure 6 A line graph of the contribution in three dimensions is given. Figure 5 and Figure 6 , the threshold of the comprehensive score SCORE is selected as N = 1, and the threshold of the contribution EP is selected as N = 3. Table 6 gives an example of a three-dimensional fundamental factor combination, as follows:

[0110] Table 6 An example of a three-dimensional fundamental factor combination

[0111]

[0112] As shown in Table 6, the three-dimensional fundamental factor combinations are channel 2 & product 2 & city 1, channel 1 & product 1 & city 1, channel 1 & product 2 & city 1, channel 1 & product 1 & city 2.

[0113] It should be noted that the order of determining the root factor combination of each dimension type is fixed, that is, first determine the one-dimensional root factor combination, then use the one-dimensional root factor combination and other root factor combinations to determine the two-dimensional root factor combination, and then use the one-dimensional root factor combination, the two-dimensional root factor combination and other root factor combinations to determine the three-dimensional root factor combination. The number of factor dimension types or the number of dimensions in the basic data set is the maximum number of times the root factor is located. The embodiment of the present invention adopts the method of gradually determining the root factor combination corresponding to each factor dimension type from low dimension to high dimension, so that after determining the root factor combination corresponding to each factor dimension type, a part of the root factor combination can be screened out, that is, the root factor combination with low root cause evaluation index value is automatically eliminated, and then when determining the root factor combination corresponding to the next factor dimension type, the number of elements in the root factor combination sequence is reduced, so as to avoid wasting computing resources, improve computing efficiency, and further improve the efficiency and accuracy of determining the root factor combination corresponding to each factor dimension type.

[0114] S260. If it is detected that the factor dimension type of the high-dimensional fundamental factor combination is the highest dimension type, determine the target fundamental factor combination that causes data fluctuations between two adjacent observation periods based on the fundamental factor combination corresponding to each factor dimension type.

[0115] Specifically, taking the three-dimensional type as an example, if it is detected that the factor dimension type of a one-dimensional fundamental factor combination is not the highest dimension type, the one-dimensional fundamental factor combination is taken as a low-dimensional fundamental factor combination, and based on the aforementioned steps and the one-dimensional fundamental factor combination, the high-dimensional fundamental factor combination at this time, that is, the two-dimensional fundamental factor combination, is determined; if it is detected that the factor dimension type of a two-dimensional fundamental factor combination is not the highest dimension type, the two-dimensional fundamental factor combination is also taken as a low-dimensional fundamental factor combination, and based on the aforementioned steps, the one-dimensional fundamental factor combination, the two-dimensional fundamental factor combination and other fundamental factor combinations, the high-dimensional fundamental factor combination at this time, that is, the three-dimensional fundamental factor combination, is determined; if it is detected that the factor dimension type of a three-dimensional fundamental factor combination is the highest dimension type, the target fundamental factor combination that causes data fluctuations between two adjacent observation periods is determined based on the one-dimensional fundamental factor combination, the two-dimensional fundamental factor combination and the three-dimensional fundamental factor combination.

[0116] On the basis of the above technical solution, "determining the target fundamental factor combination that causes data fluctuations between two adjacent observation periods based on the fundamental factor combination corresponding to each factor dimension type" may include: screening the fundamental factor combinations based on the root cause evaluation index values ​​corresponding to the fundamental factor combinations of each factor dimension type, and determining the target fundamental factor combination that causes data fluctuations between two adjacent observation periods.

[0117] Specifically, based on the pre-set root factor combination screening rules, the root factor combinations are screened using the root cause evaluation index values ​​corresponding to the root factor combinations of each factor dimension type to determine the target root factor combination that causes data fluctuations between two adjacent observation periods. The root factor combination screening rules may be, but are not limited to, retaining the root factor combination with the largest root cause evaluation index value if there is an inclusion relationship between the root factor combinations.

[0118] On the basis of the above technical scheme, "screening the root factor combinations based on the root cause evaluation index values ​​corresponding to the root factor combinations of each factor dimension type to determine the target root factor combination that causes data fluctuations between two adjacent observation periods" may include: determining the current root factor combination based on the root cause evaluation index values ​​corresponding to the low-dimensional root factor combinations of low-dimensional types and the root cause evaluation index values ​​corresponding to the high-dimensional root factor combinations that include the low-dimensional root factor combinations; deduplicating each current root factor combination to determine the target root factor combination that causes data fluctuations between two adjacent observation periods.

[0119] Specifically, if the root cause evaluation index values ​​corresponding to the low-dimensional root factor combinations of low-dimensional types are all smaller than the root cause evaluation index values ​​corresponding to the high-dimensional root factor combinations containing the low-dimensional root factor combinations, the high-dimensional root factor combinations are determined as the current root factor combinations. After the current root factor combinations are obtained by screening based on the root cause evaluation index values, the current root factor combinations can be deduplicated, that is, if there is an inclusion relationship between the root factor combinations, the root factor combinations with high factor dimension types are retained.

[0120] Exemplarily, in the i-th dimension root factor combination, the root factor combinations whose root cause evaluation index is higher than the root cause evaluation index of the one-dimensional root factor combination corresponding to the factor are further screened, and the root factor combinations with low root cause evaluation index are eliminated. Among them, i>=1 and i<=the highest dimension type corresponding to the basic data set-1. The advantage of this setting is that if the high-dimensional root cause evaluation index is lower than the one-dimensional root cause evaluation index of each of its factors, it indicates that the high-dimensional root factor combination is not the root cause, and is not as influential as the one-dimensional root factor combination, thereby improving the accuracy of determining the target root factor combination.

[0121] Specifically, for the multidimensional root cause factor combination {(W1,V1),(W2,V2),...,(Wj,Vj)},j>=2, if the root cause evaluation index values ​​of {(W1,V1),(W2,V2),...,(Wj,Vj)} are all greater than the root cause evaluation index values ​​of (Wi,Vi)(i=1,...,j), then the multidimensional root cause factor combination is retained, otherwise it is eliminated to obtain the current root factor combination. The root causes with intersections in the current root factor combination are removed to obtain the target root factor combination. For example, assuming that the one-dimensional root factor combination is (Wi,Vj) and the two-dimensional root factor combination is {(Wi,Vj),(Wm,Vn)}, the basic data set screened out by the two-dimensional root factor combination is a subset of the basic data set of the one-dimensional root factor combination. Therefore, it shows that the two-dimensional root factor combination further narrows the positioning range of the root cause corresponding to the abnormal fluctuation, which means that the two-dimensional root factor combination is closer to the root cause. At this time, (Wi, Vj) is ignored and only {(Wi, Vj), (Wm, Vn)} is retained. Specifically, if the factor in the i-th dimension root factor combination is included in the factor set in the i+1-th dimension root factor combination, then the factor is eliminated in the i-th dimension root factor combination and only retained in the i+1-th dimension root factor combination. Among them, i>=1 and i<=the highest dimension type corresponding to the basic data set -1. Table 7 gives an example of a target root factor combination, as follows:

[0122] Table 7 Examples of target root factor combinations

[0123]

[0124]

[0125] See Table 7, the target root factor combination is channel 1, product 1, channel 2 & product 2. For analysts, it is necessary to optimize the marketing strategies of channel 1, product 1, channel 2 & product 2 to prevent their cancellation rates from further deteriorating.

[0126] The technical solution of the embodiment of the present invention is as follows: if all factor dimension types are one-dimensional types, then based on each candidate root factor combination in the candidate root factor combination sequence corresponding to the one-dimensional type and the root cause evaluation index value of each candidate root factor combination, a one-dimensional root factor combination is determined, and the one-dimensional root factor combination is used as the target root factor combination that causes data fluctuations between two adjacent observation periods, thereby accurately determining the target root factor combination that causes data fluctuations between two adjacent observation periods for the candidate root factor combination sequence corresponding to the one-dimensional type. If there are two-dimensional types or multi-dimensional types in each factor dimension type, then based on the low-dimensional fundamental factor combination corresponding to the low-dimensional type and other fundamental factor combinations, the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type is determined, and based on the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type, the high-dimensional fundamental factor combination is determined; other fundamental factor combinations are fundamental factor combinations under other candidate dimension combinations except the candidate dimension combination to which the low-dimensional fundamental factor combination belongs, so that for the sequence of candidate fundamental factor combinations of multiple dimension types, the one-dimensional fundamental factor combination corresponding to the one-dimensional type is first determined, and then the two-dimensional fundamental factor combination corresponding to the two-dimensional type is determined based on the one-dimensional fundamental factor combination and other fundamental factor combinations, and then the three-dimensional fundamental factor combination corresponding to the three-dimensional type is determined based on the one-dimensional fundamental factor combination, the two-dimensional fundamental factor combination and other fundamental factor combinations, and so on, the fundamental factor combination corresponding to all factor dimension types is determined, and the determination efficiency of the fundamental factor combination corresponding to all factor dimension types is further improved. If the factor dimension type of the high-dimensional fundamental factor combination is detected to be the highest dimension type, the target fundamental factor combination that causes data fluctuations between two adjacent observation periods is determined based on the fundamental factor combination corresponding to each factor dimension type, and the determination efficiency and accuracy of the target fundamental factor combination are further improved.

[0127] The following is an embodiment of a device for determining the root cause of data fluctuations provided in an embodiment of the present invention. The device and the methods for determining the root cause of data fluctuations in the above-mentioned embodiments belong to the same inventive concept. For details not fully described in the embodiment of the device for determining the root cause of data fluctuations, reference can be made to the embodiments of the methods for determining the root cause of data fluctuations in the above-mentioned embodiments.

[0128] Embodiment 3

[0129] Figure 7 This is a schematic diagram of the structure of a data fluctuation root cause determination device provided in Embodiment 3 of the present invention. Figure 7 As shown, the device includes: an observation data acquisition module 710, a root cause evaluation index value determination module 720, a candidate root factor combination sequence determination module 730 and a target root factor combination determination module 740.

[0130] Among them, the observation data acquisition module 710 is used to obtain the first observation data corresponding to the first observation period and the second observation data corresponding to the second observation period in two adjacent observation periods, wherein the first observation data includes the first key indicator value corresponding to each candidate root factor combination, and the second observation data includes the second key indicator value corresponding to each candidate root factor combination; the root cause evaluation indicator value determination module 720 is used to determine the root cause evaluation indicator value corresponding to each candidate root factor combination based on the first key indicator value and the second key indicator value corresponding to the same candidate root factor combination; the candidate root factor combination sequence determination module 730 is used to arrange the candidate root factor combinations corresponding to each factor dimension type based on the root cause evaluation indicator value and the factor dimension type corresponding to each candidate root factor combination, and determine the candidate root factor combination sequence corresponding to each factor dimension type; the target root factor combination determination module 740 is used to determine the target root factor combination that causes data fluctuations between two adjacent observation periods based on the candidate root factor combination sequence corresponding to each factor dimension type.

[0131] The technical solution of the embodiment of the present invention is to obtain first observation data corresponding to the first observation period and second observation data corresponding to the second observation period in two adjacent observation periods, wherein the first observation data includes the first key indicator value corresponding to each candidate root factor combination, and the second observation data includes the second key indicator value corresponding to each candidate root factor combination; based on the first key indicator value and the second key indicator value corresponding to the same candidate root factor combination, determine the root cause evaluation indicator value corresponding to each candidate root factor combination, so that the data fluctuation between the first key indicator value and the second key indicator value of each candidate root factor combination in two adjacent observation periods is accurately quantified according to a unified standard, thereby improving the accuracy of determining the root cause evaluation indicator value; based on the root cause evaluation indicator value and the factor dimension type corresponding to each candidate root factor combination, arrange the candidate root factor combinations corresponding to each factor dimension type, and determine the candidate root factor combination sequence corresponding to each factor dimension type; thereby, based on the candidate root factor combination sequence corresponding to each factor dimension type, accurately and conveniently determine the target root factor combination that causes data fluctuation between two adjacent observation periods, thereby improving the efficiency and accuracy of determining the target root factor combination.

[0132] On the basis of the above technical solution, the root cause evaluation index value includes: at least one of contribution, deviation and comprehensive score; the root cause evaluation index value determination module 720 includes at least one of the following:

[0133] The contribution determination submodule is used to determine the first key indicator difference based on the first key indicator value and the second key indicator value corresponding to the same candidate fundamental factor combination; determine the second key indicator difference based on the first key indicator total value and the second key indicator total value corresponding to the candidate dimension combination to which the same candidate fundamental factor combination belongs; and determine the contribution corresponding to the same candidate fundamental factor combination by dividing the first key indicator difference and the second key indicator difference;

[0134] The deviation determination submodule is used to determine the first key indicator ratio based on the total value of the first key indicator and the first key indicator value corresponding to the same candidate fundamental factor combination; determine the second key indicator ratio based on the total value of the second key indicator and the second key indicator value corresponding to the same candidate fundamental factor combination; determine the deviation corresponding to the same candidate fundamental factor combination based on the difference between the second key indicator ratio and the first key indicator ratio and the sum of the second key indicator ratio and the first key indicator ratio;

[0135] The comprehensive score determination submodule is used to determine the comprehensive score corresponding to the same candidate fundamental factor combination by multiplying the contribution and deviation corresponding to the same candidate fundamental factor combination.

[0136] Based on the above technical solution, the target fundamental factor combination determination module 740 may include:

[0137] The first target root factor combination determination submodule is used to determine the one-dimensional root factor combination based on each candidate root factor combination in the candidate root factor combination sequence corresponding to the one-dimensional type and the root cause evaluation index value of each candidate root factor combination if all factor dimension types are one-dimensional types, and use the one-dimensional root factor combination as the target root factor combination that causes data fluctuations between two adjacent observation periods;

[0138] A high-dimensional fundamental factor combination determination submodule is used to determine the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type based on the low-dimensional fundamental factor combination corresponding to the low-dimensional type and other fundamental factor combinations if there are two-dimensional types or multi-dimensional types in each factor dimension type, and determine the high-dimensional fundamental factor combination based on the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type; other fundamental factor combinations are fundamental factor combinations under other candidate dimension combinations except the candidate dimension combinations to which the low-dimensional fundamental factor combinations belong;

[0139] The second target fundamental factor combination determination submodule is used to determine the target fundamental factor combination that causes data fluctuations between two adjacent observation periods based on the fundamental factor combination corresponding to each factor dimension type if the factor dimension type of the high-dimensional fundamental factor combination is detected to be the highest dimension type.

[0140] On the basis of the above technical solution, the first target fundamental factor combination determination submodule may include:

[0141] A maximum difference absolute value determination unit is used to determine the maximum absolute value of the difference in root cause evaluation index values ​​between adjacent candidate root factor combinations in the candidate root factor combination sequence corresponding to the one-dimensional type;

[0142] The one-dimensional fundamental factor combination determining unit is used to determine the one-dimensional fundamental factor combination based on the candidate fundamental factor combination located in the previous position in the adjacent candidate fundamental factor combinations corresponding to the maximum absolute value of the difference and each candidate factor combination before the candidate fundamental factor combination in the candidate fundamental factor combination sequence.

[0143] On the basis of the above technical solution, the maximum difference absolute value determination unit is specifically used to: determine a line graph based on the root cause evaluation index values ​​corresponding to a preset number of candidate root factor combinations in the sequence of candidate root factor combinations corresponding to the one-dimensional type; determine the maximum absolute value of the difference in root cause evaluation index values ​​between adjacent candidate root factor combinations in the sequence of candidate root factor combinations corresponding to the one-dimensional type based on the candidate root factor combination with the largest corner in the line graph.

[0144] On the basis of the above technical solution, the second target fundamental factor combination determination submodule may include:

[0145] The target root factor combination determination unit is used to screen the root factor combinations based on the root cause evaluation index values ​​corresponding to the root factor combinations of each factor dimension type, and determine the target root factor combination that causes data fluctuations between two adjacent observation periods.

[0146] On the basis of the above technical scheme, the target root factor combination determination unit is specifically used to: determine the current root factor combination based on the root cause evaluation index value corresponding to the low-dimensional root factor combination of the low-dimensional type and the root cause evaluation index value corresponding to the high-dimensional root factor combination containing the low-dimensional root factor combination; deduplicate each current root factor combination to determine the target root factor combination that causes data fluctuations between two adjacent observation periods.

[0147] The data fluctuation root cause determination device provided in the embodiment of the present invention can execute the data fluctuation root cause determination method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the data fluctuation root cause determination method.

[0148] It is worth noting that in the above-mentioned embodiment of determining the root cause of data fluctuations, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0149] Embodiment 4

[0150] Figure 8 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0151] like Figure 8 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0152] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0153] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a method for determining the root cause of data fluctuations.

[0154] In some embodiments, the data fluctuation root cause determination method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the data fluctuation root cause determination method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the data fluctuation root cause determination method in any other appropriate manner (e.g., by means of firmware).

[0155] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0156] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0157] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0158] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0159] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0160] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0161] An embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements the method for determining the root cause of data fluctuations as provided in any embodiment of the present application.

[0162] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present invention, and the programming language includes an object-oriented programming language, such as Java, Smalltalk, C++, and also includes a conventional procedural programming language, such as "C" language or similar programming language. The program code can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (for example, using an Internet service provider to connect through the Internet). The program product belongs to the same inventive concept as the data fluctuation root cause determination method disclosed in each embodiment of the present application, so it is not repeated here.

[0163] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0164] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for determining the root cause of data fluctuation, characterized in that: include: Acquire first observation data corresponding to a first observation period and second observation data corresponding to a second observation period in two adjacent observation periods, wherein the first observation data includes a first key indicator value corresponding to each candidate fundamental factor combination, and the second observation data includes a second key indicator value corresponding to each candidate fundamental factor combination; Determine the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination; Based on the root cause evaluation index value and the factor dimension type corresponding to each candidate root factor combination, the candidate root factor combinations corresponding to each factor dimension type are arranged to determine a sequence of candidate root factor combinations corresponding to each factor dimension type; Based on the sequence of candidate fundamental factor combinations corresponding to each factor dimension type, a target fundamental factor combination causing data fluctuations between the two adjacent observation periods is determined.

2. The method according to claim 1, characterized in that The root cause evaluation index value includes: at least one of contribution, deviation and comprehensive score; The determining of the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination includes at least one of the following: Determine the first key indicator difference based on the first key indicator value and the second key indicator value corresponding to the same candidate fundamental factor combination; determine the second key indicator difference based on the first key indicator total value and the second key indicator total value corresponding to the candidate dimension combination to which the same candidate fundamental factor combination belongs; determine the contribution corresponding to the same candidate fundamental factor combination based on dividing the first key indicator difference and the second key indicator difference; Determine a first key indicator ratio based on the first key indicator total value and the first key indicator value corresponding to the same candidate fundamental factor combination; determine a second key indicator ratio based on the second key indicator total value and the second key indicator value corresponding to the same candidate fundamental factor combination; determine a deviation corresponding to the same candidate fundamental factor combination based on the difference between the second key indicator ratio and the first key indicator ratio and the sum of the second key indicator ratio and the first key indicator ratio; The comprehensive score corresponding to the same candidate fundamental factor combination is determined by multiplying the contribution and deviation corresponding to the same candidate fundamental factor combination.

3. The method according to claim 1, characterized in that The determining of the target fundamental factor combination causing the data fluctuation between the two adjacent observation periods based on the sequence of candidate fundamental factor combinations corresponding to each factor dimension type includes: If all factor dimension types are one-dimensional, then based on each candidate root factor combination in the candidate root factor combination sequence corresponding to the one-dimensional type and the root cause evaluation index value of each candidate root factor combination, a one-dimensional root factor combination is determined, and the one-dimensional root factor combination is used as the target root factor combination that causes the data fluctuation between the two adjacent observation periods; If there are two-dimensional types or multi-dimensional types in each factor dimension type, then based on the low-dimensional fundamental factor combination corresponding to the low-dimensional type and other fundamental factor combinations, determine the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type, and based on the sequence of candidate fundamental factor combinations corresponding to the high-dimensional type, determine the high-dimensional fundamental factor combination; the other fundamental factor combinations are the fundamental factor combinations under other candidate dimension combinations except the candidate dimension combination to which the low-dimensional fundamental factor combination belongs; If it is detected that the factor dimension type of the high-dimensional fundamental factor combination is the highest dimension type, then the target fundamental factor combination that causes the data fluctuation between the two adjacent observation periods is determined based on the fundamental factor combination corresponding to each factor dimension type.

4. The method according to claim 3, characterized in that The determining of the one-dimensional root factor combination based on each candidate root factor combination in the sequence of candidate root factor combinations corresponding to the one-dimensional type and the root cause evaluation index value of each candidate root factor combination comprises: Determine the maximum absolute value of the difference between the root cause evaluation index values ​​of adjacent candidate root factor combinations in the candidate root factor combination sequence corresponding to the one-dimensional type; A one-dimensional root factor combination is determined based on a previous candidate root factor combination in adjacent candidate root factor combinations corresponding to the maximum absolute value of the difference and each candidate factor combination before the candidate root factor combination in the candidate root factor combination sequence.

5. The method according to claim 4, characterized in that The maximum absolute value of the difference between the root cause evaluation index values ​​of adjacent candidate root factor combinations in the candidate root factor combination sequence corresponding to the one-dimensional type is determined, including: Determine a line chart based on root cause evaluation index values ​​corresponding to a preset number of candidate root factor combinations in a sequence of candidate root factor combinations corresponding to the one-dimensional type; Based on the candidate root factor combination with the largest corner in the line graph, the maximum absolute value of the difference in root cause evaluation index values ​​between adjacent candidate root factor combinations in the candidate root factor combination sequence corresponding to the one-dimensional type is determined.

6. The method according to claim 3, characterized in that The determining of the target fundamental factor combination causing the data fluctuation between the two adjacent observation periods based on the fundamental factor combination corresponding to each factor dimension type includes: The root factor combinations are screened based on the root cause evaluation index values ​​corresponding to the root factor combinations of each factor dimension type, and the target root factor combinations that cause the data fluctuations between the two adjacent observation periods are determined.

7. The method according to claim 6, characterized in that The root cause evaluation index value corresponding to the root factor combination of each factor dimension type is used to screen the root factor combination to determine the target root factor combination that causes data fluctuations between the two adjacent observation periods, including: Determine the current root factor combination based on the root cause evaluation index value corresponding to the low-dimensional root factor combination of the low-dimensional type and the root cause evaluation index value corresponding to the high-dimensional root factor combination including the low-dimensional root factor combination; De-duplicate each current fundamental factor combination to determine the target fundamental factor combination that causes data fluctuations between the two adjacent observation periods.

8. A device for determining the root cause of data fluctuation, characterized in that: The device comprises: An observation data acquisition module, used to acquire first observation data corresponding to a first observation period and second observation data corresponding to a second observation period in two adjacent observation periods, wherein the first observation data includes a first key indicator value corresponding to each candidate root factor combination, and the second observation data includes a second key indicator value corresponding to each candidate root factor combination; A root cause evaluation index value determination module, used to determine the root cause evaluation index value corresponding to each candidate root factor combination based on the first key index value and the second key index value corresponding to the same candidate root factor combination; A candidate root factor combination sequence determination module is used to arrange the candidate root factor combinations corresponding to each factor dimension type based on the root cause evaluation index value and the factor dimension type corresponding to each candidate root factor combination, and determine the candidate root factor combination sequence corresponding to each factor dimension type; The target fundamental factor combination determination module is used to determine the target fundamental factor combination that causes data fluctuations between the two adjacent observation periods based on the sequence of candidate fundamental factor combinations corresponding to each factor dimension type.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining the root cause of data fluctuation as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for determining the root cause of data fluctuation as described in any one of claims 1 to 7 is implemented.