Index fluctuation analysis method, device, computer storage medium and system

Abstract

The embodiment of the present application discloses an indicator fluctuation analysis method, device, computer storage medium and system, the method comprising: obtaining a massive amount of original data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy comprises at least a drill-down analysis strategy and / or an allocation analysis strategy; when the drill-down analysis strategy is used to perform an indicator fluctuation analysis on the massive amount of original data, a first influence of at least one analysis dimension on the overall indicator value is determined, and an indicator fluctuation analysis result is determined according to the first influence; when the allocation analysis strategy is used to perform an indicator fluctuation analysis on the massive amount of original data, a second influence of an allocation result of at least one upstream node on a change in an allocation value of a target node is determined, and an indicator fluctuation analysis result is determined according to the second influence; the indicator fluctuation analysis result is stored, and / or the indicator fluctuation analysis result is displayed in a preset manner.

Description

Technical Field

[0001] The present application relates to the technical field of massive data processing, and in particular to an indicator fluctuation analysis method, device, computer storage medium and system. Background Art

[0002] In the business analysis or financial analysis of enterprises, abnormal fluctuations in certain important indicators are often found. Data analysts need to trace the source of these abnormal fluctuations and find the root causes, so as to provide necessary support for the formulation and adjustment of management decision-making strategies, prevent problems from happening, and maintain the healthy growth of related businesses.

[0003] In the existing technology, in current business analysis or financial analysis, the detection of abnormal fluctuations in indicators mainly relies on experience and simple logic to analyze the causes of abnormal fluctuations, which lacks accuracy and scientificity; and some relatively mature attribution analysis models are usually used in qualitative analysis scenarios of user behavior, which are difficult to transplant to complex quantitative analysis scenarios of business indicators; at the same time, there is still a lack of effective solutions for massive data in actual analysis, which seriously affects the timeliness and accuracy of the analysis. Summary of the invention

[0004] In view of this, the present application proposes an indicator fluctuation analysis method, device, computer storage medium and system, which can not only improve the analysis efficiency and accuracy of indicator fluctuation attribution analysis, but also improve the timeliness of analysis under massive data.

[0005] The technical solution of this application is implemented as follows:

[0006] In a first aspect, an embodiment of the present application provides an indicator fluctuation analysis method, the method comprising:

[0007] Obtaining a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy at least includes a drill-down analysis strategy and / or an apportionment analysis strategy;

[0008] When the indicator fluctuation analysis is performed on the massive raw data using the drill-down analysis strategy, a first influence degree of at least one analysis dimension on the overall indicator value is determined, and the indicator fluctuation analysis result is determined according to the first influence degree;

[0009] When using the apportionment analysis strategy to perform an index fluctuation analysis on a large amount of raw data, determine a second influence degree of the apportionment result of at least one upstream node on the change of the apportionment value of the target node, and determine the index fluctuation analysis result according to the second influence degree;

[0010] The indicator fluctuation analysis results are stored, and / or the indicator fluctuation analysis results are displayed in a preset manner.

[0011] In a second aspect, an embodiment of the present application provides an indicator fluctuation analysis device, the indicator fluctuation analysis device includes an acquisition unit, a determination unit and an output unit; wherein:

[0012] An acquisition unit configured to acquire a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy at least includes a drill-down analysis strategy and / or an apportionment analysis strategy;

[0013] a determination unit configured to determine a first influence degree of at least one analysis dimension on the overall indicator value when performing indicator fluctuation analysis on the massive raw data using a drill-down analysis strategy, and determine an indicator fluctuation analysis result according to the first influence degree;

[0014] The determination unit is further configured to determine a second influence degree of the allocation result of at least one upstream node on the change of the allocation value of the target node when the massive raw data is analyzed for the index fluctuation using the allocation type analysis strategy, and determine the index fluctuation analysis result according to the second influence degree;

[0015] The output unit is configured to store the indicator fluctuation analysis results and / or display the indicator fluctuation analysis results in a preset manner.

[0016] In a third aspect, an embodiment of the present application provides an indicator fluctuation analysis device, the indicator fluctuation analysis device comprising a memory and a processor; wherein:

[0017] A memory for storing computer programs that can be run on the processor;

[0018] A processor is used to execute the method described in the first aspect when running a computer program.

[0019] In a fourth aspect, an embodiment of the present application provides a computer storage medium, which stores a computer program. When the computer program is executed by a processor, the method described in the first aspect is implemented.

[0020] In a fifth aspect, an embodiment of the present application provides an indicator fluctuation analysis system, which indicator fluctuation analysis system at least includes the indicator fluctuation analysis device as described in the second aspect or the third aspect.

[0021] An indicator fluctuation analysis method, device, computer storage medium and system provided in the embodiments of the present application obtain massive original data to be analyzed and corresponding analysis strategies; wherein the analysis strategies at least include drill-down analysis strategies and / or allocation analysis strategies; when the drill-down analysis strategy is used to perform indicator fluctuation analysis on massive original data, a first influence of at least one analysis dimension on the overall indicator value is determined, and the indicator fluctuation analysis result is determined based on the first influence; when the allocation analysis strategy is used to perform indicator fluctuation analysis on massive original data, a second influence of the allocation result of at least one upstream node on the change of the allocation value of the target node is determined, and the indicator fluctuation analysis result is determined based on the second influence; the indicator fluctuation analysis result is stored, and / or the indicator fluctuation analysis result is displayed in a preset manner. In this way, for the indicator fluctuation attribution analysis scenario under massive data, in order to solve the bottlenecks of the existing technology that relies solely on experience but lacks scientificity, accuracy and timeliness, the present application uses a drill-down analysis strategy and / or an allocation analysis strategy to perform indicator fluctuation analysis on massive raw data, which can not only improve the analysis efficiency and accuracy of the indicator fluctuation attribution analysis, but also improve the analysis timeliness under massive data. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flow chart of an indicator fluctuation analysis method provided in an embodiment of the present application;

[0023] Figure 2 A schematic diagram of a flow chart of another indicator fluctuation analysis method provided in an embodiment of the present application;

[0024] Figure 3 A detailed flow chart of an indicator fluctuation analysis method provided in an embodiment of the present application;

[0025] Figure 4 A flowchart of another indicator fluctuation analysis method provided in an embodiment of the present application;

[0026] Figure 5 A partial schematic diagram of a flow network of a distributed indicator provided in an embodiment of the present application;

[0027] Figure 6 A detailed flowchart of another indicator fluctuation analysis method provided in an embodiment of the present application;

[0028] Figure 7 A schematic diagram of the composition structure of an indicator fluctuation analysis device provided in an embodiment of the present application;

[0029] Figure 8 A schematic diagram of the hardware structure of an indicator fluctuation analysis device provided in an embodiment of the present application;

[0030] Fig. 9A schematic diagram of the structure of an electronic device provided in an embodiment of the present application;

[0031] Fig.10 A schematic diagram of the composition structure of an indicator fluctuation analysis system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to enable a more detailed understanding of the features and technical contents of the embodiments of the present application, the implementation of the embodiments of the present application is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present application.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0034] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. In addition, the terms "first\second\third" involved in the embodiments of the present application are only used to distinguish similar objects and do not represent a specific order for the objects. It is understood that "first\second\third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0035] With the development of information technology, all walks of life generate a large amount of data every day during the operation process, such as some report data. It is a direction of great research value to detect whether these data have abnormal fluctuations and analyze the reasons for abnormal fluctuations. Specifically, in the business analysis or financial analysis of enterprises, it is often found that some important indicators have abnormal fluctuations. Data analysts need to trace these abnormal fluctuations and find the root causes, so as to provide necessary support for the formulation and adjustment of management decision-making strategies, prevent problems from happening, and maintain the healthy growth of related businesses.

[0036] In practical applications, this type of analysis is mostly based on experience and feeling, and is carried out with simple logic. For example, the "elimination" method, "contribution rate" method and other logics can be used to drill down on each dimension to evaluate the influence of each dimension. In addition, in the e-commerce field and other fields, there are already some relatively mature attribution analysis models in analysis scenarios such as user conversion, such as heuristic attribution models such as the first / last interaction model, linear attribution model, time decay attribution model, U-shaped attribution model, and machine learning algorithm attribution models such as the Logistic model, Cox model, probability model, and Markov model. However, these models are generally used for qualitative analysis scenarios of user behavior and are difficult to transplant to complex business indicator quantitative attribution analysis scenarios. In addition, in the era of big data, the raw data of business analysis can easily reach tens of millions or even hundreds of millions of rows, and the analysis process needs to be completed as real-time as possible. This places higher demands on computing resources, especially memory and algorithm efficiency.

[0037] In current business analysis or financial analysis, there is no established methodological system for attributing the causes of indicator fluctuations. On the one hand, the results obtained based on experience and simple logic lack accuracy and scientificity; on the other hand, various models for user behavior are difficult to transplant to complex business indicator quantitative analysis scenarios; at the same time, there is also a lack of effective solutions for massive data in actual analysis, which seriously affects the timeliness and accuracy of analysis.

[0038] Based on this, an embodiment of the present application provides an indicator fluctuation analysis method, the basic idea of ​​which is: obtaining massive original data to be analyzed and the corresponding analysis strategy; wherein the analysis strategy includes at least a drill-down analysis strategy and / or an allocation analysis strategy; when using the drill-down analysis strategy to perform indicator fluctuation analysis on massive original data, determine the first influence of at least one analysis dimension on the overall indicator value, and determine the indicator fluctuation analysis result based on the first influence; when using the allocation analysis strategy to perform indicator fluctuation analysis on massive original data, determine the second influence of the allocation result of at least one upstream node on the change of the target node allocation value, and determine the indicator fluctuation analysis result based on the second influence; store the indicator fluctuation analysis result, and / or display the indicator fluctuation analysis result in a preset manner. In this way, for the indicator fluctuation attribution analysis scenario under massive data, in order to solve the bottlenecks of the existing technology that relies solely on experience but lacks scientificity, accuracy and timeliness, the present application uses a drill-down analysis strategy and / or an allocation analysis strategy to perform indicator fluctuation analysis on massive raw data, which can not only improve the analysis efficiency and accuracy of the indicator fluctuation attribution analysis, but also improve the analysis timeliness under massive data.

[0039] The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0040] In one embodiment of the present application, see Figure 1 , which shows a flow chart of an indicator fluctuation analysis method provided by an embodiment of the present application. Figure 1 As shown, the method may include:

[0041] S101: Acquire a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy at least includes a drill-down analysis strategy and / or an apportionment analysis strategy.

[0042] It should be noted that the indicator fluctuation analysis method described in the embodiment of the present application is applicable to scenarios with massive data, especially business analysis and financial analysis scenarios under massive data.

[0043] It should also be noted that the index fluctuation analysis method described in the embodiment of the present application is applied to an index fluctuation analysis device, or an electronic device integrated with the device. In the embodiment of the present application, the electronic device can be implemented in various forms, for example, the electronic device can be such as a smart phone, a tablet computer, a laptop computer, a PDA, a portable media player (PMP), a navigation device, a wearable device, a smart bracelet, a desktop computer, a server, and the like.

[0044] In the embodiment of the present application, for massive raw data, in order to analyze the causes of abnormal fluctuations in certain important indicators, two analysis strategies are proposed, namely, a drill-down analysis strategy and an apportionment analysis strategy.

[0045] Among them, for the drill-down analysis strategy, the dimensional drill-down method can be used for analysis. Drill-down can refer to expanding the next layer of data from the current data. For example, if the lower level of the classification of a certain data is the product name, then drilling down means expanding from the classification list to the product name list. For the allocation analysis strategy, the impact of each link in the data flow process on the fluctuation of indicators can be analyzed. For example, the procurement and sales expenses of an e-commerce company are allocated from the highest-level business group to each first-level department according to certain rules, and then to each second-level department, third-level department, ..., and finally to each product, such as the stock keeping unit (SKU).

[0046] In this way, for massive amounts of raw data that are suitable for drill-down analysis strategies, the drill-down analysis strategies can be used to perform indicator fluctuation analysis; for massive amounts of raw data that are suitable for allocation analysis strategies, the allocation analysis strategies can be used to perform indicator fluctuation analysis.

[0047] S102: When performing an indicator fluctuation analysis on a large amount of raw data using a drill-down analysis strategy, determine a first influence of at least one analysis dimension on an overall indicator value, and determine an indicator fluctuation analysis result according to the first influence.

[0048] S103: When performing an index fluctuation analysis on a large amount of raw data using an apportionment analysis strategy, determine a second influence of an apportionment result of at least one upstream node on a change in an apportionment value of a target node, and determine an index fluctuation analysis result based on the second influence.

[0049] It should be noted that the fluctuation of indicators can be expressed in various forms, such as year-on-year growth rate and month-on-month growth rate. For year-on-year and month-on-month, year-on-year refers to the comparison between the current period and the same period (or "same period last year"), and month-on-month refers to the comparison between the current period and the previous period (or "previous statistical cycle"). For example, the comparison between July 2005 and July 2004 is called year-on-year; the comparison with the previous statistical period, such as the comparison between July 2005 and June 2005, is called month-on-month.

[0050] In the embodiment of the present application, whether the year-on-year increase or the month-on-month increase is used, the principle of the index fluctuation analysis is similar, and the only difference is the numerical value. In a specific example, the embodiment of the present application can perform the index fluctuation analysis in the form of year-on-year increase.

[0051] In some embodiments, for S102, when the indicator fluctuation analysis is performed on the massive raw data using the drill-down analysis strategy, the determining of the first influence of at least one analysis dimension on the overall indicator value may include:

[0052] Determine the indicators to be analyzed for massive amounts of raw data;

[0053] Drill down the dimension of the indicator to be analyzed to determine the scope dimension corresponding to the indicator to be analyzed, and the scope dimension includes at least one analysis dimension;

[0054] Calculate the overall indicator value of the indicator to be analyzed in the scope dimension;

[0055] According to the overall indicator value, a first influence degree of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension is calculated.

[0056] It should be noted that when massive amounts of raw data are suitable for indicator fluctuation analysis by drilling down dimensions, a drill-down analysis strategy can be used. Specifically, after determining the indicators to be analyzed of the massive amounts of raw data, the indicators to be analyzed can be drilled down in dimensions to determine the range dimensions corresponding to the indicators to be analyzed and at least one analysis dimension included in the range dimensions; then the overall indicator value of the indicator to be analyzed in the range dimension is calculated in the form of year-on-year growth rate, and then the first degree of influence of the indicators to be analyzed on the overall indicator value in at least one analysis dimension is calculated based on the overall indicator value.

[0057] It should also be noted that the dimension drill-down may include the drill-down of the scope dimension and the drill-down of the analysis dimension.

[0058] Thus, the dimension drilling down of the indicator to be analyzed may include:

[0059] Drill down the scope dimension of the indicator to be analyzed to determine the scope dimension corresponding to the indicator to be analyzed;

[0060] Drill down the analysis dimension of the indicator to be analyzed based on the scope dimension to determine at least one analysis dimension included in the scope dimension.

[0061] That is to say, taking the scope dimension as department and the analysis dimension as category as an example, the drill-down of the scope dimension can be the drill-down of the department dimension, that is, to perform analysis and calculation within each department; thus, the drill-down of the analysis dimension is performed on the basis of the drill-down of the department dimension, and the first influence of each analysis dimension on the overall indicator value is calculated. For example, drilling down the commodity category dimension within a department, that is, calculating the influence of each category on the change of the department's overall analysis indicator (such as transaction amount).

[0062] In some embodiments, for S103, when the index fluctuation analysis is performed on the massive raw data using the apportionment analysis strategy, the second degree of influence of the apportionment result of at least one upstream node on the change of the apportionment value of the target node may include:

[0063] Determine the preliminary results of the impact of the change in the apportionment value of at least one leaf node;

[0064] Perform redundant pruning on the preliminary results of the impact of the change in the apportioned value of at least one leaf node, and obtain preliminary results of the impact of the to-be-apportioned value of the root node, preliminary results of the numerator impact of nodes at each layer, and preliminary results of the denominator impact;

[0065] Preprocess the preliminary results of the impact of the value to be apportioned of the root node, the preliminary results of the numerator impact and the preliminary results of the denominator impact of the nodes at each layer, and obtain the impact result of the value to be apportioned of the root node and the numerator impact result and the denominator impact result of each upstream node corresponding to the target node;

[0066] The influence degree result of the to-be-apportioned value of the root node and the numerator influence degree result and denominator influence degree result of each upstream node corresponding to the target node are determined as the second influence degree.

[0067] It should be noted that when massive raw data is suitable for analyzing the impact of each link in the data flow process on the fluctuation of indicators, the apportionment analysis strategy can be used at this time, specifically to build a data flow network diagram. Taking the tree diagram as an example, it can include several leaf nodes, the highest level is the root node, and each raw data in the data mart is a parent node-child node pair and an apportionment rule between the two nodes. When the apportionment value of the leaf node fluctuates greatly, it is necessary to determine which factors (value to be apportioned, numerator or denominator) of which upstream nodes are mainly responsible for the leaf node (i.e. the target node).

[0068] It should also be noted that at least one leaf node may include a target node and each upstream node of the target node, and preprocessing may include mapping processing and normalization processing. In this way, after determining the preliminary results of the influence of the change in the apportioned value of at least one leaf node, the preliminary results of the influence of the change in the apportioned value of the at least one leaf node may be subjected to processing operations such as redundant pruning, mapping, and normalization, thereby obtaining the influence results of the to-be-apportioned value of the root node and the numerator influence results and denominator influence results of each upstream node corresponding to the target node, that is, the second influence of the apportionment result of at least one upstream node on the change in the apportioned value of the target node.

[0069] S104: storing the indicator fluctuation analysis result, and / or displaying the indicator fluctuation analysis result in a preset manner.

[0070] In an embodiment of the present application, after determining the results of the indicator fluctuation analysis, the indicator analysis results can be output to the client for visual display. For example, for a drill-down analysis strategy, a preset number of indicator analysis results (such as the Top10 or Bottom10 of the first influence, etc.) can be displayed, or a red light or yellow light warning can be given if the negative impact exceeds the preset threshold; for an apportionment analysis strategy, it can be output to a web page for display in an interactive manner such as a data flow diagram, a tree diagram, a weighted directed graph, etc. In addition, persistent storage can also be performed for the indicator analysis results, such as storage in a data mart, a distributed file system (Hadoop Distributed File System, HDFS) or a database such as ClickHouse.

[0071] The present embodiment provides an index fluctuation analysis method, by obtaining a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy includes at least a drill-down analysis strategy and / or an apportionment analysis strategy; when the drill-down analysis strategy is used to perform index fluctuation analysis on the massive raw data, the first influence of at least one analysis dimension on the overall index value is determined, and the index fluctuation analysis result is determined according to the first influence; when the apportionment analysis strategy is used to perform index fluctuation analysis on the massive raw data, the second influence of the apportionment result of at least one upstream node on the change of the apportionment value of the target node is determined, and the index fluctuation analysis result is determined according to the second influence; the index fluctuation analysis result is stored, and / or the index fluctuation analysis result is displayed in a preset manner. In this way, for the index fluctuation attribution analysis scenario under massive data, in order to solve the bottlenecks of the prior art that only relies on experience and lacks scientificity, accuracy and timeliness, the present application uses the drill-down analysis strategy and / or the apportionment analysis strategy to perform index fluctuation analysis on the massive raw data, which can not only improve the analysis efficiency and accuracy during the index fluctuation attribution analysis, but also improve the analysis timeliness under massive data.

[0072] In another embodiment of the present application, see Figure 2 , which shows a flow chart of another indicator fluctuation analysis method provided by an embodiment of the present application. Figure 2 As shown, the method may include:

[0073] S201: Determine indicators to be analyzed of the massive original data according to the massive original data to be analyzed.

[0074] It should be noted that the analysis indicators of the embodiment of the present application involve various aspects such as finance and operation, and the amount of data is extremely large and is uniformly stored in a data mart. Here, the data mart can be described in a table form, that is, these massive raw data are uniformly stored in a data mart table.

[0075] It should also be noted that, since different indicators may come from different sources and have different data volumes, their refresh times are bound to be different. For example, some monthly data can be run and written into the market within a week, while some require more than two weeks. Therefore, in some embodiments, the method may also include: using a real-time monitoring module to monitor changes in the data mart in real time; if there is an update to the indicator data, obtaining a large amount of raw data to be analyzed and the corresponding indicator to be analyzed, so as to subsequently perform the impact calculation of the indicator.

[0076] S202: Drill down the dimension of the indicator to be analyzed to determine the range dimension corresponding to the indicator to be analyzed, and the range dimension includes at least one analysis dimension.

[0077] It should be noted that the dimension drill-down here may include the scope dimension drill-down and the analysis dimension drill-down. Thus, by drilling down the scope dimension, the scope dimension can be determined; and on this basis, by drilling down the analysis dimension, at least one analysis dimension can be determined.

[0078] S203: Calculate the overall indicator value of the indicator to be analyzed in the range dimension.

[0079] In the embodiment of the present application, the indicator type may include a first indicator type, a second indicator type and a third indicator type; wherein different indicator types have different corresponding overall value calculation models. Therefore, in some embodiments, for S203, the calculation of the overall indicator value of the indicator to be analyzed in the range dimension may include:

[0080] Identify the indicator type corresponding to the indicator to be analyzed;

[0081] The overall value calculation model corresponding to the indicator to be analyzed is determined according to the identified indicator type, and the overall value calculation model is used to calculate the overall indicator value of the indicator to be analyzed in the range dimension.

[0082] Here, the indicator type may include a first indicator type, a second indicator type, and a third indicator type, and the overall value calculation model may include a first overall value calculation model, a second overall value calculation model, and a third overall value calculation model. Then in a specific example, the calculation of the overall indicator value of the indicator to be analyzed in the range dimension may include:

[0083] If the identified indicator type is the first indicator type, the overall indicator value of the indicator to be analyzed in the range dimension is calculated using the first overall value calculation model;

[0084] If the identified indicator type is the second indicator type, the overall indicator value of the indicator to be analyzed in the range dimension is calculated using the second overall value calculation model;

[0085] If the identified indicator type is the third indicator type, the third overall value calculation model is used to calculate the overall indicator value of the indicator to be analyzed in the range dimension.

[0086] It should be noted that, in a more specific example, the first indicator type can be an atomic indicator, the second indicator type can be a ratio-based derivative indicator, and the third indicator type can be a non-ratio-based derivative indicator. For atomic indicators, that is, indicators that can be directly obtained from the original data and meet additivity, such as transaction amount, market costs, etc.; for ratio-based derivative indicators, it can be the quotient of two atomic indicators and is a ratio, for example: market rate, its value = market cost / net income; for non-ratio-based derivative indicators, it can be obtained by simple calculation from atomic indicators, for example: unit price, its value = transaction amount / transaction sales volume.

[0087] In addition, in the embodiment of the present application, the indicators can also be classified according to their positive and negative nature. In this case, the indicator types can include two types, which can be specifically divided into positive indicators and negative indicators. Among them, positive indicators are indicators with larger values, the better, such as transaction amount; negative indicators are indicators with smaller values, the better, such as market fees.

[0088] It can be understood that, taking the change as the year-on-year increase as an example, the embodiment of the present application can respectively take the above dimensions (including the scope dimension and analysis dimension of the drill-down) and the summary data of the atomic indicator in this period and the same period last year. Among them, if the data in this period or the same period is empty, then the calculation process of the indicator can be terminated.

[0089] In a possible implementation, taking the first indicator type, namely the atomic indicator (eg, transaction amount) as an example, the year-on-year comparison is: current period value / same period value-1.

[0090] Here, assuming that the overall value of this period is X, and the overall value of the same period is X′; then the first overall value calculation model, that is, the overall year-on-year calculation formula is:

[0091] In another possible implementation, taking the second indicator type, i.e., ratio-based derivative indicator (eg, market rate, whose value = market expense / net income) as an example, the year-on-year comparison is: current period value - same period value.

[0092] Here, it is assumed that the overall value is During the same period Among them, N and D are the total values ​​of the additive atomic indicators in this period, and N′ and D′ are their total values ​​in the same period respectively; then the second total value calculation model, that is, the total year-on-year It should be noted that if the denominator is 0 during the calculation process, the calculation process of the indicator will be terminated.

[0093] In another possible implementation, taking the third indicator type, i.e., non-ratio derivative indicator (e.g., unit price, whose value = transaction amount / transaction sales volume) as an example, the year-on-year comparison is: current period value / same period value-1.

[0094] Here, it is assumed that the overall value is During the same period Among them, N and D are the total values ​​of the additive atomic indicators in this period, and N′ and D′ are their total values ​​in the same period respectively; then the third total value calculation model, that is, the total year-on-year It should be noted that if the denominator is 0 during the calculation process, the calculation process of the indicator will be terminated.

[0095] In this way, after determining the indicator type, the corresponding overall value calculation model (such as the first overall value calculation model, the second overall value calculation model or the third overall value calculation model) can be selected to calculate the overall indicator value of the indicator to be analyzed in the range dimension.

[0096] S204: Calculate, according to the overall indicator value, a first influence of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension.

[0097] It should be noted that the embodiment of the present application also designs an influence calculation model, which can be used to calculate the first influence of each of at least one analysis dimension on the overall index value. In some embodiments, the calculation of the first influence of each of the indicators to be analyzed on the overall index value in at least one analysis dimension according to the overall index value may include:

[0098] Determine the impact calculation model of the indicator to be analyzed;

[0099] According to the overall indicator value, the influence degree calculation model is used to calculate the first influence degree of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension.

[0100] In the embodiment of the present application, the influence calculation model may include an elimination method model, a control variable method model and a proportion method model. Among them, for the elimination method model, its basic idea is to study the influence by assuming elimination; for the control variable method model, its basic idea is to study the influence by assuming that certain data values ​​are unchanged; for the proportion method model, its basic idea is to study the influence by considering the proportion, which has the following properties: the sum of the influence of each part is equal to the overall year-on-year value.

[0101] It should also be noted that the embodiments of the present application design a switch mechanism. Specifically, the elimination method model, the control variable method model, and the proportional method model are connected to three switching switches; thus, in some embodiments, the determination of the influence degree calculation model of the indicator to be analyzed may include: by controlling the conduction state of the switching switch, determining one or more of the elimination method model, the control variable method model, and the proportional method model as the influence degree calculation model of the indicator to be analyzed.

[0102] That is to say, the embodiments of the present application can select a model, such as any one of the elimination method model, the control variable method model and the proportional method model, by controlling the conduction state of the switching switch according to the needs of the user; it can also select multiple models, such as the elimination method model and the control variable method model, the control variable method model and the proportional method model, the elimination method model and the proportional method model, or multiple combinations of the elimination method model, the control variable method model and the proportional method model, which are not specifically limited here.

[0103] Further, when multiple models are selected, in some embodiments, the using the influence calculation model to calculate the first influence of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension may include:

[0104] When multiple models among the elimination method model, the control variable method model and the proportion method model are determined as the influence degree calculation models of the indicator to be analyzed, the multiple models are used to respectively calculate the first sub-influence of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension;

[0105] A weighted calculation is performed on the first sub-influences obtained by using multiple models to obtain the first influence of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension.

[0106] It should be noted that if at least two models are selected from the elimination method model, the control variable method model and the proportion method model, then it is necessary to perform a weighted calculation on the first sub-influence obtained using multiple models to obtain the first influence of the indicator to be analyzed on the overall indicator value in at least one analysis dimension.

[0107] It should also be noted that for the elimination method model, the control variable method model and the proportion method model, these three influence calculation models have different influence calculation models for different indicator types. Therefore, in some embodiments, the method may also include:

[0108] Identify the indicator type corresponding to the indicator to be analyzed;

[0109] According to the identified indicator type, the elimination method model, control variable method model and proportion method model corresponding to the indicator to be analyzed are determined; among which, the elimination method model, control variable method model and proportion method model corresponding to different indicator types are different.

[0110] In a possible implementation, for the first indicator type (i.e., atomic indicator), assuming that the analysis dimension has a total of m enumeration values ​​(for example, if the scope dimension is department and the analysis dimension is category, it means that there are a total of m categories in a department), the value of the enumeration value i (i=1, 2, ..., m) is currently x i , and x′ i ,So At this time, the influence of the enumeration value i on the overall is calculated as follows:

[0111] (a) The first influence degree is calculated using the elimination method model as shown below:

[0112]

[0113] (b) The first influence degree is calculated using the control variable method model as shown below:

[0114]

[0115] (c) The first influence degree is calculated using the proportional method model as follows:

[0116]

[0117] In another possible implementation, for the second indicator type (i.e., ratio-based derived indicator), assuming that the analysis dimension has a total of m enumerated values, the value of the enumerated value i (i=1, 2, ..., m) is currently During the same period So At this time, the influence of the enumeration value i on the overall is calculated as follows:

[0118] (a) The first influence degree is calculated using the elimination method model as shown below:

[0119]

[0120] (b) The first influence degree is calculated using the control variable method model as shown below:

[0121]

[0122] (c) The first influence degree is calculated using the proportional method model as follows:

[0123]

[0124] In another possible implementation, for the third indicator type (i.e., non-ratio derived indicator), assuming that the analysis dimension has a total of m enumeration values, the value of the enumeration value i (i=1, 2, ..., m) is currently During the same period So At this time, the influence of the enumeration value i on the overall is calculated as follows:

[0125] (a) The first influence degree is calculated using the elimination method model as shown below:

[0126]

[0127] (b) The first influence degree is calculated using the control variable method model as shown below:

[0128]

[0129] (c) The first influence degree is calculated using the proportional method model as follows:

[0130]

[0131] In addition, the atomic index can be expressed as the ratio of two atomic indexes (i.e. ), the embodiment of the present application can directly adopt the above logic processing. For other derived indicators, such as X=Y±Z or If we consider Y±Z and / or P±Q as derived indices (where Y, Z, P, and Q are all atomic indices), then we can also consider Y±Z and / or P±Q as atomic indices and process them according to the above logic.

[0132] S205: Determine the indicator fluctuation analysis result according to the first influence degree.

[0133] It should be noted that after obtaining the first impact degree, the indicators to be analyzed can also be sorted according to the first impact degree according to the positivity and business needs of the indicators to be analyzed. In some embodiments, determining the indicator fluctuation analysis result according to the first impact degree may include: arranging the first impact degree according to a first preset method, and generating the indicator fluctuation analysis result according to the arrangement result; wherein the first preset method includes an ascending method or a descending method.

[0134] That is to say, according to the positivity and negativity of the indicator to be analyzed and the business needs, the enumeration values ​​of each dimension are sorted in a preset manner (ascending or descending order) according to the first impact degree to determine the indicator fluctuation analysis results; then the indicator fluctuation analysis results are visualized and / or persistently stored according to actual needs.

[0135] For example, for business analysis and financial analysis scenarios under massive data, the present application embodiment designs a drill-down indicator fluctuation attribution analysis system to analyze the causes of indicator fluctuations in a drill-down dimension manner. Figure 3 , which shows a detailed flow chart of an indicator fluctuation analysis method provided by an embodiment of the present application. Figure 3 As shown, the detailed process may include:

[0136] S301: Perform real-time detection on the data mart.

[0137] S302: Determine the indicator to be analyzed.

[0138] S303: Identify the indicator type.

[0139] S304: Drill down on the dimension.

[0140] S305: Data reading and aggregation.

[0141] S306: Calculate overall value.

[0142] S307: First influence calculation.

[0143] S308: Perform sorting and integration to determine the results of the indicator fluctuation analysis.

[0144] S309: Persistently store the indicator fluctuation analysis results.

[0145] S310: Visually display the indicator fluctuation analysis results.

[0146] It should be noted that in actual analysis, the fluctuation analysis of indicators can be in various forms such as year-on-year growth rate, month-on-month growth rate, etc. The following uses the year-on-year growth rate as an example to describe the working principle of the drill-down indicator fluctuation attribution analysis system. Figure 3 The detailed process is as follows:

[0147] Step 1: Real-time monitoring step. The analysis indicators of the embodiment of the present application involve various aspects such as finance and operation, and the amount of data is extremely large, which is uniformly stored in the data mart. Since the sources of different indicators may be different and the amount of data varies greatly, there must be differences in their refresh time. For example, some monthly data can be run and written to the market within a week, while some require more than two weeks. Therefore, a real-time monitoring module can be used to monitor changes in the data source of the market. If there is an update to the indicator data, the impact calculation of the indicator is performed.

[0148] Step 2: Identify the indicator type. According to the calculation logic, the indicators involved here can be divided into three types:

[0149] (1) Atomic indicators, that is, indicators that can be directly obtained from the original data and meet the additivity, such as transaction amount, market cost, etc.; (2) Ratio-type derived indicators, which are the quotient of two atomic indicators and are ratios, such as: market rate, its value = market cost / net income; (3) Non-ratio-type derived indicators, which are obtained by simple calculation from atomic indicators, such as: unit price, its value = transaction amount / transaction volume. In addition, according to the positive and negative nature, the indicator type can be divided into positive indicators and negative indicators; among them, positive indicators are indicators whose values ​​are larger, the better, such as transaction amount; negative indicators are indicators whose values ​​are smaller, the better, such as market cost.

[0150] Step 3: Drill down in dimensions. This can be divided into drilling down in scope dimensions and drilling down in analysis dimensions.

[0151] ① Drill down on the scope dimension. For example, drill down on the department dimension, that is, perform analysis and calculation within each department;

[0152] ② Drill down the analysis dimension, that is, calculate the impact of each enumeration value of the analysis dimension on the overall change based on the first step. For example, drill down the commodity category dimension within a department, that is, calculate the impact of each category on the change of the department's overall analysis indicators (such as transaction amount).

[0153] It should also be noted that the analysis dimensions can be categories, gender, date of birth, etc. Among them, taking gender as an example, its enumeration value can be: male and female; taking category as an example, its enumeration value can be: food, tops, pants, skirts, sports shoes, cloth shoes, etc. In other words, enumeration is a fixed type, and an enumeration corresponds to several defined values, which are called enumeration values. For example: suppose the enumeration of week is defined, because it is week, it will have seven values, specifically Monday to Sunday, and each of these seven can be called an enumeration value.

[0154] Step 4: Data reading and aggregation. After determining the indicators to be analyzed, the scope dimensions of the drill-down, and the analysis dimensions, read the data of the relevant dimensions and atomic indicators from the data mart, and aggregate and sum the atomic indicator values ​​according to the dimensions, for example, SUM (transaction amount) GROUP BY department, category; output the aggregated data to the next step (i.e., the impact calculation module). Taking the change as the year-on-year increase as an example, the aggregated data of the above dimensions and atomic indicators for the current period and the same period last year can be taken respectively. If the data for the current period or the same period is empty, the calculation process of the indicator is terminated.

[0155] Step 5: Calculate the overall value. Here, you can calculate the year-on-year growth of the internal analysis indicators of the enumeration values ​​of each range dimension. For different indicator types, the year-on-year calculation method is as follows:

[0156] ① Atomic indicators (e.g. transaction amount), the year-on-year change is: current period value / same period value - 1. If the overall value of the current period is X, and the overall value of the same period is X', then the overall year-on-year change is

[0157] ② Ratio-based derivative indicators (e.g. market rate, value = market cost / net income), the year-on-year comparison is: current period value - same period value. During the same period Among them, N and D are the overall values ​​of the additive atomic indicators in this period, and N' and D' are their overall values ​​in the same period respectively; then the overall year-on-year If the denominator is 0 during the calculation, the calculation process of the indicator can be terminated.

[0158] ③ Non-ratio derivative indicators (for example: unit price, its value = transaction amount / transaction volume), its year-on-year comparison is: current period value / same period value - 1. During the same period The meanings of the symbols are the same as above. If the denominator is 0 during the calculation, the calculation process of the indicator can be terminated.

[0159] Step 6: Calculate the first impact of each enumeration value in the analysis dimension.

[0160] Here, the embodiment of the present application designs three general influence calculation models: (1) the elimination method model, whose basic idea is to study the influence by assuming elimination; (2) the control variable method model, whose basic idea is to study the influence by assuming that certain data values ​​are unchanged; (3) the proportion method model, whose basic idea is to study the influence by considering the proportion, and it has the following properties: the sum of the influence of each part is equal to the overall year-on-year value. Among them, these three models have different influence calculation methods for the three indicator types. In addition, the embodiment of the present application also designs a switch mechanism, and the user can choose one model or multiple models for weighted combination. The following describes the influence calculation methods of each model for the three indicator types.

[0161] ① Atomic indicators. Assuming that the analysis dimension has a total of m enumeration values ​​(for example, if the scope dimension is department and the analysis dimension is category, it means that there are a total of m categories in a department), the value of enumeration value i (i = 1, 2, ..., m) is x i , and x′ i ,have At this time, the influence of the enumeration value i on the overall is calculated as follows:

[0162] The first influence degree is calculated using the elimination method model, as shown in the above formula (1).

[0163] The first influence degree is calculated using the control variable method model, as shown in the above formula (2).

[0164] The first influence degree is calculated using the proportionality model, as shown in the above formula (3).

[0165] ② Ratio-type derived indicators. Assume that the analysis dimension has a total of m enumeration values, and the value of enumeration value i (i = 1, 2, ..., m) is During the same period So At this time, the influence of the enumeration value i on the overall is calculated as follows:

[0166] The first influence degree is calculated using the elimination method model, as shown in the above formula (4).

[0167] The first influence degree is calculated using the control variable method model, as shown in the above formula (5).

[0168] The first influence degree is calculated using the proportionality model, as shown in the above formula (6).

[0169] ③ Non-ratio derived indicators. Assume that the analysis dimension has m enumeration values, and the value of enumeration value i (i = 1, 2, ..., m) is During the same period So At this time, the influence of the enumeration value i on the overall is calculated as follows:

[0170] The first influence degree is calculated using the elimination method model, as shown in the above formula (7).

[0171] The first influence degree is calculated using the control variable method model, as shown in the above formula (8).

[0172] The first influence degree is calculated using the proportionality model, as shown in the above formula (9).

[0173] In addition, the atomic index can be expressed as the ratio of two atomic indexes (i.e. ), the embodiment of the present application can directly adopt the above logic processing. For other derived indicators, such as X=Y±Z or If we consider Y±Z and / or P±Q as derived indices (where Y, Z, P, and Q are all atomic indices), then we can also consider Y±Z and / or P±Q as atomic indices and process them according to the above logic.

[0174] Step 7: When calculating the influence of multiple models in the elimination method model, control variable method model and proportion method model, the comprehensive influence of the enumeration value of each dimension (such as category value) on the overall indicator value (such as the overall transaction amount of the department) can be weighted to determine the first influence.

[0175] Step 8: According to the positivity and business needs of the indicators to be analyzed, sort the enumeration values ​​of each dimension in a preset manner (ascending or descending order) according to the comprehensive impact.

[0176] Step 9: Output and storage. According to specific needs, the above indicator fluctuation analysis results are output to the front end (such as the client) for visual display; for example, display the Top10 or Bottom10 enumeration values ​​of the impact, or give a red light or yellow light warning if the negative impact exceeds the set threshold. At the same time, the above indicator fluctuation analysis result data can also be persistently stored; for example, stored in a data mart or a database such as HDFS or ClickHouse.

[0177] In another embodiment of the present application, see Figure 4 , which shows a flow chart of another indicator fluctuation analysis method provided by an embodiment of the present application. Figure 4 As shown, the method may include:

[0178] S401: Determine a preliminary result of the impact of a change in the apportionment value of at least one leaf node based on the massive amount of original data to be analyzed.

[0179] It should be noted that the analysis indicators of the embodiments of the present application may also involve various aspects such as finance and operation. In real analysis scenarios, massive amounts of apportionment relationship data may be involved, which requires extremely high memory and computing efficiency. The main method is to use layer-by-layer calculation to solve the problem of large data volume and incomplete information.

[0180] by Figure 5 As an example, it shows a partial schematic diagram of a flow network of a distributed indicator provided in an embodiment of the present application. Figure 5 In the example, the leaf nodes at the bottom (the nodes represented by solid circles) are the 0th layer, and the layers above are the 1st layer, the 2nd layer, and so on, until the root node. Figure 5 The fifth layer (nodes represented by dotted hollow circles) is shown in Figure 1. Each piece of raw data in the data mart is a parent node-child node pair and the allocation rule between the two nodes, that is, Figure 5 The part within the dotted circle shown. Figure 5 As shown, the allocation relationship is complicated. There may be multiple allocation paths upstream of a leaf node (from a root node to a leaf node is an allocation path), and each original data has only local information, and the leaf node or root node of the allocation path cannot be directly obtained.

[0181] In some embodiments, determining a preliminary result of the influence of a change in the apportioned value of at least one leaf node may include:

[0182] Set the initialization value of l to 1;

[0183] Read the relevant data of the lth layer. If the read data is not empty, perform a preliminary calculation of the impact of the relevant data of the lth layer to obtain the preliminary results of the impact of the changes in the apportionment values ​​of several leaf nodes included in the lth layer;

[0184] Execute plus-one calculation on l, and return to execute the step of reading relevant data of the lth layer until the read data is empty, so as to obtain a preliminary result of the influence of the change of the apportionment value of at least one leaf node.

[0185] In a specific example, the preliminary calculation of the influence of the relevant data of the first layer to obtain the preliminary influence results of the changes in the apportionment values ​​of several leaf nodes included in the first layer may include:

[0186] The impact calculation model of the value to be apportioned is used to determine the preliminary results of the impact of the value to be apportioned of the changes in the apportioned values ​​of several leaf nodes included in the first layer;

[0187] The molecular influence calculation model is used to determine the preliminary results of the molecular influence of the changes in the apportioned values ​​of several leaf nodes included in the first layer;

[0188] The denominator influence calculation model is used to determine the preliminary results of the denominator influence of the changes in the apportioned values ​​of several leaf nodes included in the lth layer.

[0189] In the embodiment of the present application, for the calculation of the preliminary results of the influence of each node in each path of the first layer, assuming a certain allocation relationship k, the current period values ​​of the parent node to be allocated, the numerator and the denominator are respectively and The values ​​for the same period are and The corresponding leaf node i is from the path where the relationship k is located (it may be one or more, see Figure 5 ) The apportioned values ​​for this period and the same period are and The apportioned values ​​of the leaf node obtained from all paths in this period and the same period are V0 and V′0 respectively; then the total apportioned value change rate of leaf node i is η0=V0 / V′0-1, and the overall contribution rate of apportionment relationship k to the change of apportioned value of leaf node i is The preliminary results of the influence of various factors of the allocation relationship k on the change of the allocation value of the leaf node i are as follows:

[0190] (a) For the preliminary result of the impact of the value to be apportioned, the impact calculation model of the value to be apportioned shown in formula (10) can be used for calculation.

[0191]

[0192] (b) For the preliminary results of molecular influence, the molecular influence calculation model shown in formula (11) can be used for calculation.

[0193]

[0194] (c) For the preliminary result of the denominator influence, the denominator influence calculation model shown in formula (12) can be used for calculation.

[0195]

[0196] S402: Redundancy is pruned for the preliminary results of the impact of the change in the apportioned value of at least one leaf node to obtain the preliminary results of the impact of the to-be-apportioned value of the root node, the preliminary results of the numerator impact of the nodes at each layer, and the preliminary results of the denominator impact.

[0197] It should be noted that the to-be-apportioned values ​​of the nodes at each level below the root node are completely dependent on the to-be-apportioned values, numerators, and denominators of their upstream nodes. These can be regarded as redundant items, and therefore can be pruned. At this time, only the preliminary results of the influence of the to-be-apportioned value of the root node and the preliminary results of the influence of the numerators and denominators of the nodes at each level are retained.

[0198] S403: Preprocess the preliminary results of the influence of the value to be apportioned of the root node, the preliminary results of the numerator influence and the preliminary results of the denominator influence of the nodes at each layer, and obtain the results of the influence of the value to be apportioned of the root node and the numerator influence and the denominator influence of each upstream node corresponding to the target node.

[0199] It should be noted that the preprocessing here may include: mapping processing and normalization processing. In some embodiments, the preprocessing of the preliminary results of the impact of the to-be-apportioned value of the root node, the preliminary results of the numerator impact of the nodes at each layer, and the preliminary results of the denominator impact may include:

[0200] Mapping is performed on the preliminary results of the impact of the to-be-allocated value of the root node, the preliminary results of the numerator impact of the nodes at each layer, and the preliminary results of the denominator impact to obtain the output results after mapping;

[0201] The output results after mapping are normalized to obtain the influence result of the to-be-apportioned value of the root node and the numerator influence result and denominator influence result of each upstream node corresponding to the target node.

[0202] It should also be noted that the influence result of the to-be-allocated value of the root node obtained after normalization and the numerator influence result and denominator influence result of each upstream node corresponding to the target node all have a value range of 0 to 1.

[0203] S404: Determine the impact result of the to-be-allocated value of the root node and the numerator impact result and denominator impact result of each upstream node corresponding to the target node as the second impact, and determine the indicator fluctuation analysis result according to the second impact.

[0204] It should be noted that, after obtaining the second influence degree, determining the indicator fluctuation analysis result according to the second influence degree may include: arranging the second influence degrees according to a second preset manner, and generating the indicator fluctuation analysis result according to the arrangement result.

[0205] In an embodiment of the present application, the second preset method may include a data flow graph method, a tree graph method, or a weighted directed graph method. In other words, the indicator fluctuation analysis results can be visualized and / or persistently stored in an interactive manner such as a data flow graph method, a tree graph method, or a weighted directed graph method.

[0206] For example, for business analysis and financial analysis scenarios under massive data, the present application embodiment designs an apportioned indicator fluctuation attribution analysis system to analyze the impact of each link in the data flow process on indicator fluctuations. Figure 6 , which shows a detailed flow chart of another indicator fluctuation analysis method provided by an embodiment of the present application. Figure 6As shown, the detailed process may include:

[0207] S601: Initialization: l=1.

[0208] S602: Read the relevant data of the first layer.

[0209] S603: Determine whether the data is empty.

[0210] S604: If the judgment result is no, then calculate and store the preliminary result of the influence degree of the lth layer.

[0211] S605: l=l+1.

[0212] S606: If the judgment result is yes, perform redundancy pruning to obtain a preliminary result of the influence.

[0213] S607: Mapping the preliminary results of the influence.

[0214] S608: Normalize the preliminary results of the influence.

[0215] S609: Determine the index fluctuation analysis result according to the second influence degree of each node factor.

[0216] S610: Persistently store the indicator fluctuation analysis results.

[0217] S611: Visualize the indicator fluctuation analysis results.

[0218] It should be noted that in business analysis and financial analysis, there is a type of indicator that is allocated. For example, the purchasing and sales expenses of an e-commerce company are allocated from the highest-level business group to each first-level department according to certain rules, and then to each second-level department, third-level department, and finally to each SKU. Figure 5 For example, take the bottom leaf node (such as SKU, Figure 5 The solid nodes in the figure are the 0th level, and the levels above are the 1st level, the 2nd level, and so on, until the root node (e.g., business group, Figure 5 Each piece of raw data in the data mart is a parent node-child node pair and the allocation rule between the two nodes, that is, Figure 5 The part within the dotted circle in the middle. The essence of the allocation rule is the allocation ratio, which can be composed of a numerator and a denominator (due to the nature of the business, the numerator and the denominator have clear meanings and need to be provided separately, and cannot be directly replaced by a ratio). The index allocation value obtained by the child node from an allocation relationship = the index to be allocated value of the parent node × numerator / denominator. When the allocation value of the leaf node fluctuates greatly, it is necessary to know which factor (to be allocated value, numerator or denominator) of which upstream link (i.e. node) is mainly responsible.

[0219] In actual analysis scenarios, a large amount of allocation relationship data may be involved, which requires extremely high memory and computing efficiency. In addition, the allocation relationship is complicated (such as Figure 5 As shown in the figure, there may be multiple allocation paths upstream of a leaf node (one allocation path is from a root node to a leaf node), and each piece of original data only has local information, and the leaf node or root node of the allocation path cannot be directly obtained. The system uses a layer-by-layer association method to solve the above problems, and performs redundant pruning, mapping, and normalization on the preliminary results of the impact degree, which can ensure the scientificity and standardization of the final output results. The following uses the year-on-year increase as an example to describe the working principle of the allocation-type indicator fluctuation attribution analysis system. Combined with Figure 6 The detailed process is as follows:

[0220] Step 1: Starting from the first layer, the raw data of each layer is processed in turn. The processing process of the lth layer (l = 1, 2, 3, ...) is as follows:

[0221] First, read the current and same period data related to the lth layer from the data mart. Adopting the step-by-step association method, for each allocation relationship data, only retain its lth layer (parent node layer) node ID, to-be-allocated value, numerator, denominator, l-1 layer (child node layer) node ID (only used to mark the allocation relationship: an allocation relationship is uniquely determined by the parent node ID and child node ID), and the corresponding 0th layer leaf node ID, the allocation value obtained from the current allocation relationship, and the total allocation value obtained, and discard the relevant data of the intermediate layers (i.e., 1st to l-2nd layers), among which the total allocation value of each node in the 0th layer can be placed in the memory for reuse. If the read data is empty, it means that the highest level has been exceeded and the second step is entered; otherwise, continue to read the current and same period data related to the lth layer from the data mart.

[0222] Secondly, the influence of each node in each path of the lth layer is preliminarily calculated. Assuming a certain allocation relationship k, the parent node's to-be-allocated value, the current period value of the numerator and the denominator are and The values ​​for the same period are and The corresponding leaf node i obtains the apportioned value of the current period and the same period from the path (possibly one or more) where the relationship k is located. and The apportionment values ​​obtained by the leaf node from all paths in this period and the same period are V0 and V′0 respectively. Then the total apportionment value change rate of leaf node i is η0=V0 / V′0-1, and the overall contribution rate of apportionment relationship k to the apportionment value change of leaf node i is The preliminary results of the influence of various factors of the allocation relationship k on the change of the allocation value of the leaf node i are as follows:

[0223] The preliminary result of the impact of the value to be apportioned can be calculated using the above formula (10).

[0224] The preliminary results of the molecular influence can be calculated using the above formula (11).

[0225] The preliminary result of the denominator influence can be calculated using the above formula (12).

[0226] Step 2: Redundancy pruning. After obtaining the preliminary results of the influence of each leaf node, the level of its root node can be determined. Among them, the value to be apportioned of each layer of nodes below the root node depends entirely on the value to be apportioned, the numerator and the denominator of its upstream node, which can be regarded as redundant items. Therefore, the preliminary results related to these values ​​to be apportioned are pruned, leaving only the value to be apportioned of the root node and the preliminary results related to the numerator and denominator of each layer of nodes.

[0227] Step 3: Mapping. Since the goal of the analysis is to trace the root cause of abnormal fluctuations, the results that have a counter-effect on the actual fluctuations can be ignored, so the linear rectified linear unit (ReLU) function can be used to map the preliminary results of the impact.

[0228] Step 4: Normalization. For ease of comparison, the results after ReLU mapping are normalized with leaf nodes as units to obtain the influence of the numerator and denominator of each upstream node and the to-be-apportioned value of the root node on the fluctuation of the leaf node index. It should be noted that the value range of the influence of each factor is [0,1].

[0229] Step 5: Output and storage. According to specific needs, the indicator fluctuation analysis results obtained above are output to the front end (such as the client) for visual display; for example, they are output to the web page in an interactive way such as a data flow diagram, a tree diagram, a weighted directed graph, etc. At the same time, the indicator fluctuation analysis result data obtained above can also be persistently stored; for example, stored in a data mart or a database such as HDFS or ClickHouse.

[0230] Through the above embodiments, the specific implementation of the above embodiments is elaborated in detail. It can be seen that through the technical scheme of the above embodiments, on the one hand, for the business analysis and financial analysis scenarios under massive data, a drill-down indicator fluctuation attribution analysis system can be designed. By introducing a real-time monitoring mechanism, an influence degree calculation module composed of elimination method, control variable method and proportion method is constructed, and different logical calculations are designed for the influence of the drill-down dimension for different indicator types such as atomic indicators, ratio-type derivative indicators and non-ratio-type derivative indicators, and a switch switching mechanism is added so that users can choose one model or a weighted combination of multiple models; on the other hand, for the business analysis and financial analysis scenarios under massive data, an apportionment indicator fluctuation attribution analysis system can also be designed. The layer-by-layer calculation method is used to solve the problems of large data volume and incomplete information. Through preliminary calculation, redundant pruning, ReLU mapping and normalization, the influence of each upstream link factor on its fluctuation is obtained; thus solving the bottlenecks of the existing technology that only relies on experience but lacks a systematic and scientific method system, low analysis efficiency and accuracy, and lack of timeliness under massive data.

[0231] In yet another embodiment of the present application, see Figure 7 , which shows a schematic diagram of the composition structure of an indicator fluctuation analysis device 70 provided in an embodiment of the present application. Figure 7 As shown, the index fluctuation analysis device 70 may include: an acquisition unit 701, a determination unit 702 and an output unit 703; wherein,

[0232] The acquisition unit 701 is configured to acquire a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy at least includes a drill-down analysis strategy and / or an apportionment analysis strategy;

[0233] The determination unit 702 is configured to determine a first influence degree of at least one analysis dimension on the overall indicator value when performing indicator fluctuation analysis on the massive raw data using a drill-down analysis strategy, and determine an indicator fluctuation analysis result according to the first influence degree;

[0234] The determination unit 702 is further configured to determine a second influence degree of the allocation result of at least one upstream node on the change of the allocation value of the target node when the index fluctuation analysis is performed on the massive raw data using the allocation type analysis strategy, and determine the index fluctuation analysis result according to the second influence degree;

[0235] The output unit 703 is configured to store the indicator fluctuation analysis result and / or display the indicator fluctuation analysis result in a preset manner.

[0236] In some embodiments, see Figure 7 , the index fluctuation analysis device 70 may further include a calculation unit 704;

[0237] The determination unit 702 is further configured to determine the to-be-analyzed indicator of the massive raw data when the indicator fluctuation analysis is performed on the massive raw data using the drill-down analysis strategy; and to perform dimension drilling on the to-be-analyzed indicator to determine the range dimension corresponding to the to-be-analyzed indicator, wherein the range dimension includes at least one analysis dimension;

[0238] The calculation unit 704 is configured to calculate the overall indicator value of the indicator to be analyzed in the range dimension; and calculate the first influence of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension according to the overall indicator value.

[0239] In some embodiments, the dimension drill-down includes: drill-down of the scope dimension and drill-down of the analysis dimension.

[0240] In some embodiments, the calculation unit 704 is also configured to identify the indicator type corresponding to the indicator to be analyzed; and determine the overall value calculation model corresponding to the indicator to be analyzed according to the identified indicator type, and use the overall value calculation model to calculate the overall indicator value of the indicator to be analyzed in the range dimension.

[0241] In some embodiments, the indicator type includes a first indicator type, a second indicator type and a third indicator type, and the overall value calculation model includes a first overall value calculation model, a second overall value calculation model and a third overall value calculation model; accordingly, the calculation unit 704 is also configured to, if the identified indicator type is the first indicator type, use the first overall value calculation model to calculate the overall indicator value of the indicator to be analyzed on the range dimension; or, if the identified indicator type is the second indicator type, use the second overall value calculation model to calculate the overall indicator value of the indicator to be analyzed on the range dimension; or, if the identified indicator type is the third indicator type, use the third overall value calculation model to calculate the overall indicator value of the indicator to be analyzed on the range dimension.

[0242] In some embodiments, the determination unit 702 is further configured to determine an influence degree calculation model of the indicator to be analyzed;

[0243] The calculation unit 704 is further configured to calculate, according to the overall indicator value, a first influence of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension by using the influence calculation model.

[0244] In some embodiments, the influence calculation model includes an elimination method model, a control variable method model and a proportional method model, and the elimination method model, the control variable method model and the proportional method model are correspondingly connected to three switching switches; accordingly, the determination unit 702 is also configured to control the conduction state of the switching switch to determine one or more models among the elimination method model, the control variable method model and the proportional method model as the influence calculation model of the indicator to be analyzed.

[0245] In some embodiments, the determination unit 702 is further configured to identify the indicator type corresponding to the indicator to be analyzed; and determine the elimination method model, control variable method model and proportional method model corresponding to the indicator to be analyzed according to the identified indicator type; wherein the elimination method model, control variable method model and proportional method model corresponding to different indicator types are different.

[0246] In some embodiments, the calculation unit 704 is also configured to, when multiple models among the elimination method model, the control variable method model and the proportion method model are determined as influence calculation models for the indicator to be analyzed, use multiple models to respectively calculate the first sub-influence of the indicator to be analyzed on the overall indicator value in at least one analysis dimension; and perform weighted calculation on the first sub-influence obtained using multiple models to obtain the first influence of the indicator to be analyzed on the overall indicator value in at least one analysis dimension.

[0247] In some embodiments, see Figure 7 The indicator fluctuation analysis device 70 may further include an arrangement unit 705, which is configured to arrange the first influence degrees according to a first preset method and generate an indicator fluctuation analysis result according to the arrangement result; wherein the first preset method includes an ascending method or a descending method.

[0248] In some embodiments, the determination unit 702 is further configured to determine a preliminary result of the influence of the change in the apportioned value of at least one leaf node when an apportioned analysis strategy is used to perform an indicator fluctuation analysis on massive raw data; and to perform redundant pruning on the preliminary result of the influence of the change in the apportioned value of at least one leaf node to obtain a preliminary result of the influence of the value to be apportioned of the root node, a preliminary result of the numerator influence of the nodes at each layer, and a preliminary result of the denominator influence; and to pre-process the preliminary result of the influence of the value to be apportioned of the root node, the preliminary result of the numerator influence of the nodes at each layer, and the preliminary result of the denominator influence to obtain the result of the influence of the value to be apportioned of the root node, and the numerator influence result and denominator influence result of each upstream node corresponding to the target node; and to determine the result of the influence of the value to be apportioned of the root node, and the numerator influence result and denominator influence result of each upstream node corresponding to the target node as the second influence.

[0249] In some embodiments, the calculation unit 704 is further configured to set the initialization value of l to 1; and read the relevant data of the lth layer. If the read data is not empty, perform a preliminary calculation of the influence of the relevant data of the lth layer to obtain a preliminary result of the influence of the changes in the apportionment values ​​of several leaf nodes included in the lth layer; and perform a plus-one calculation on l, return to execute the step of reading the relevant data of the lth layer, until the read data is empty, so as to obtain a preliminary result of the influence of the changes in the apportionment value of at least one leaf node.

[0250] In some embodiments, the calculation unit 704 is further configured to determine the preliminary results of the influence of the to-be-apportioned values ​​of the changes in the apportioned values ​​of several leaf nodes included in the first layer by using the to-be-apportioned values ​​influence calculation model; and to determine the preliminary results of the numerator influence of the changes in the apportioned values ​​of several leaf nodes included in the first layer by using the numerator influence calculation model; and to determine the preliminary results of the denominator influence of the changes in the apportioned values ​​of several leaf nodes included in the first layer by using the denominator influence calculation model.

[0251] In some embodiments, preprocessing includes: mapping processing and normalization processing; and the influence result of the root node's apportioned value obtained after normalization processing and the numerator influence result and denominator influence result of each upstream node corresponding to the target node are all in the range of 0 to 1.

[0252] In some embodiments, the arrangement unit 705 is further configured to arrange the second influence degree according to a second preset method, and generate an indicator fluctuation analysis result according to the arrangement result; wherein the second preset method includes a data flow graph method, a tree graph method or a weighted directed graph method.

[0253] It can be understood that in this embodiment, a "unit" can be a part of a circuit, a part of a processor, a part of a program or software, etc., and of course it can also be a module, or it can be non-modular. Moreover, the components in this embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional module.

[0254] If the integrated unit is implemented in the form of a software function module and is not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) or a processor to perform all or part of the steps of the method described in this embodiment. The aforementioned storage medium includes: U disk, mobile hard disk, read only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0255] Therefore, this embodiment provides a computer storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the method in any one of the above embodiments are implemented.

[0256] Based on the composition of the index fluctuation analysis device 70 and the computer storage medium, see Figure 8 , which shows a hardware structure diagram of the index fluctuation analysis device provided in the embodiment of the present application. Figure 8 As shown, the indicator fluctuation analysis device 70 may include: a communication interface 801, a memory 802 and a processor 803; each component is coupled together via a bus system 804. It is understood that the bus system 804 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 804 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, Figure 8 Various buses are labeled as bus system 804. Among them, the communication interface 801 is used for receiving and sending signals in the process of sending and receiving information between other external network elements;

[0257] A memory 802, used to store a computer program that can be run on the processor 803;

[0258] The processor 803 is configured to execute, when running the computer program:

[0259] Obtaining a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy at least includes a drill-down analysis strategy and / or an apportionment analysis strategy;

[0260] When the indicator fluctuation analysis is performed on the massive raw data using the drill-down analysis strategy, a first influence degree of at least one analysis dimension on the overall indicator value is determined, and an indicator fluctuation analysis result is determined according to the first influence degree;

[0261] When the index fluctuation analysis is performed on the massive raw data using the apportionment analysis strategy, a second influence degree of the apportionment result of at least one upstream node on the change of the apportionment value of the target node is determined, and the index fluctuation analysis result is determined according to the second influence degree;

[0262] The indicator fluctuation analysis result is stored, and / or the indicator fluctuation analysis result is displayed in a preset manner.

[0263] It can be understood that the memory 802 in the embodiment of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus random access memory (DRRAM). The memory 802 of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0264] The processor 803 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 803. The above processor 803 may be a general processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor are combined and executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 802, and the processor 803 reads the information in the memory 802 and completes the steps of the above method in combination with its hardware.

[0265] It is understood that the embodiments described herein can be implemented with hardware, software, firmware, middleware, microcode or a combination thereof. Wherein, for hardware implementation, the processing unit can be implemented in one or more application specific integrated circuits (Application Specific Integrated Circuits, ASIC), digital signal processors (Digital Signal Processing, DSP), digital signal processing devices (DSP Device, DSPD), field programmable gate arrays (Field-Programmable Gate Array, FPGA), general processors, controllers, microcontrollers, microprocessors, etc. for performing other electronic units or combinations of the functions described in the present application. For software implementation, the technology described herein can be implemented by modules (such as processes, functions, etc.) that perform the functions described herein. The software code can be stored in a memory and executed by a processor. The memory can be implemented in a processor or outside a processor.

[0266] Optionally, as another embodiment, the processor 803 is further configured to execute the steps of the method in any one of the aforementioned embodiments when running the computer program.

[0267] Based on the composition and structure of the above-mentioned index fluctuation analysis device 70, see Fig. 9 , which shows a schematic diagram of the composition structure of an electronic device provided by an embodiment of the present application. Fig. 9 As shown, the electronic device 90 may at least include the indicator fluctuation analysis device 70 described in any one of the aforementioned embodiments.

[0268] Based on the composition and structure of the above-mentioned index fluctuation analysis device 70, see Fig.10 , which shows a schematic diagram of the composition structure of an indicator fluctuation analysis system provided by an embodiment of the present application. Fig.10 As shown, the indicator fluctuation analysis system 100 may at least include the indicator fluctuation analysis device 70 described in any one of the aforementioned embodiments.

[0269] In the embodiment of the present application, the index fluctuation analysis system 100 can be a drill-down index fluctuation attribution analysis system, or an apportionment index fluctuation attribution analysis system. In this way, for the index fluctuation attribution analysis scenario under massive data, the present application uses the drill-down analysis strategy and / or the apportionment analysis strategy to perform index fluctuation analysis on massive raw data, which can not only improve the analysis efficiency and accuracy during the index fluctuation attribution analysis, but also improve the analysis timeliness under massive data.

[0270] It should be noted that, in this application, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0271] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0272] The methods disclosed in several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0273] The features disclosed in several product embodiments provided in this application can be arbitrarily combined without conflict to obtain new product embodiments.

[0274] The features disclosed in several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0275] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for analyzing index fluctuations, characterized in that: The method comprises: Obtaining a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy at least includes a drill-down analysis strategy and / or an apportionment analysis strategy; When the indicator fluctuation analysis is performed on the massive raw data using the drill-down analysis strategy, a first influence degree of at least one analysis dimension on the overall indicator value is determined, and an indicator fluctuation analysis result is determined according to the first influence degree; When the index fluctuation analysis is performed on the massive raw data using the apportionment analysis strategy, the second influence of the apportionment result of at least one upstream node on the change of the apportionment value of the target node is determined, and the index fluctuation analysis result is determined according to the second influence; wherein, the second influence of the apportionment result of at least one upstream node on the change of the apportionment value of the target node is determined, including: constructing a data flow network diagram, the data flow network diagram including at least one leaf node; determining a preliminary result of the influence of the change of the apportionment value of at least one leaf node; and performing an analysis on the preliminary result of the influence of the change of the apportionment value of at least one leaf node. Prune redundant rows to obtain preliminary results of the influence of the value to be apportioned of the root node, preliminary results of the numerator influence of nodes at each layer, and preliminary results of the denominator influence; preprocess the preliminary results of the influence of the value to be apportioned of the root node, preliminary results of the numerator influence of nodes at each layer, and preliminary results of the denominator influence to obtain the influence result of the value to be apportioned of the root node, and the numerator influence result and denominator influence result of each upstream node corresponding to the target node; determine the influence result of the value to be apportioned of the root node, and the numerator influence result and denominator influence result of each upstream node corresponding to the target node as the second influence; The indicator fluctuation analysis result is stored, and / or the indicator fluctuation analysis result is displayed in a preset manner.

2. The method according to claim 1, characterized in that When the indicator fluctuation analysis is performed on the massive raw data using the drill-down analysis strategy, determining a first influence of at least one drill-down dimension on the overall indicator value includes: Determining indicators to be analyzed of the massive amount of raw data; Drilling down the dimension of the indicator to be analyzed to determine the range dimension corresponding to the indicator to be analyzed, wherein the range dimension includes at least one analysis dimension; Calculate the overall indicator value of the indicator to be analyzed in the range dimension; According to the overall indicator value, a first influence degree of each of the indicators to be analyzed on the overall indicator value in the at least one analysis dimension is calculated.

3. The method according to claim 2, characterized in that The dimension drill-down includes: drill-down of scope dimension and drill-down of analysis dimension.

4. The method according to claim 2, characterized in that: The calculating the overall indicator value of the indicator to be analyzed in the range dimension includes: Identify the indicator type corresponding to the indicator to be analyzed; The overall value calculation model corresponding to the indicator to be analyzed is determined according to the identified indicator type, and the overall value calculation model is used to calculate the overall indicator value of the indicator to be analyzed in the range dimension.

5. The method according to claim 4, characterized in that The indicator types include a first indicator type, a second indicator type and a third indicator type, and the overall value calculation model includes a first overall value calculation model, a second overall value calculation model and a third overall value calculation model; The calculating the overall indicator value of the indicator to be analyzed in the range dimension includes: If the identified indicator type is the first indicator type, the first overall value calculation model is used to calculate the overall indicator value of the indicator to be analyzed in the range dimension; If the identified indicator type is the second indicator type, the second overall value calculation model is used to calculate the overall indicator value of the indicator to be analyzed in the range dimension; If the identified indicator type is the third indicator type, the third overall value calculation model is used to calculate the overall indicator value of the indicator to be analyzed in the range dimension.

6. The method according to claim 2, characterized in that The calculating, according to the overall indicator value, a first influence degree of each of the indicators to be analyzed on the overall indicator value in the at least one analysis dimension includes: Determine an influence calculation model for the indicator to be analyzed; and calculate, according to the overall indicator value, using the influence calculation model, a first influence of each of the indicators to be analyzed on the overall indicator value in at least one analysis dimension.

7. The method according to claim 6, characterized in that The influence degree calculation model includes an elimination method model, a control variable method model and a proportional method model, and the elimination method model, the control variable method model and the proportional method model are connected to three switching switches correspondingly; The determining of the influence calculation model of the indicator to be analyzed includes: By controlling the conduction state of the switching switch, one or more models among the elimination method model, the control variable method model and the proportion method model are determined as the influence degree calculation model of the indicator to be analyzed.

8. The method according to claim 7, characterized in that The method further comprises: Identify the indicator type corresponding to the indicator to be analyzed; According to the identified indicator type, the elimination method model, the control variable method model and the proportion method model corresponding to the indicator to be analyzed are determined; wherein the elimination method model, the control variable method model and the proportion method model corresponding to different indicator types are different.

9. The method according to claim 7, characterized in that: The using the influence calculation model to calculate the first influence of each of the indicators to be analyzed on the overall indicator value in the at least one analysis dimension includes: When multiple models among the elimination method model, the control variable method model and the proportion method model are determined as the influence calculation models of the indicator to be analyzed, the multiple models are used to respectively calculate the first sub-influence of each of the indicators to be analyzed on the overall indicator value in the at least one analysis dimension; A weighted calculation is performed on the first sub-influences obtained by using the multiple models to obtain a first influence of each of the indicators to be analyzed on the overall indicator value in the at least one analysis dimension.

10. The method according to any one of claims 1 to 9, characterized in that: The determining of an indicator fluctuation analysis result according to the first influence degree includes: The first influence degrees are arranged according to a first preset method, and the indicator fluctuation analysis result is generated according to the arrangement result; wherein the first preset method includes an ascending method or a descending method.

11. The method according to claim 1, characterized in that The determining of the preliminary result of the influence of the change in the apportionment value of at least one leaf node includes: Set the initialization value of l to 1; Read the relevant data of the lth layer. If the read data is not empty, perform a preliminary calculation of the impact of the relevant data of the lth layer to obtain a preliminary result of the impact of the change in the apportionment value of several leaf nodes included in the lth layer; Perform a plus-one calculation on l, and return to execute the step of reading the relevant data of the lth layer until the read data is empty, so as to obtain a preliminary result of the influence of the change in the apportionment value of the at least one leaf node.

12. The method according to claim 11, characterized in that The preliminary calculation of the impact of the relevant data of the first layer is performed to obtain the preliminary results of the impact of the changes in the apportionment values ​​of several leaf nodes included in the first layer, including: Determine preliminary results of the impact of the to-be-apportioned values ​​of the changes in the apportioned values ​​of the several leaf nodes included in the first layer by using the to-be-apportioned value impact calculation model; Determine the preliminary results of the molecular influence of the changes in the apportioned values ​​of the several leaf nodes included in the first layer by using the molecular influence calculation model; The denominator influence calculation model is used to determine the preliminary results of the denominator influence of the changes in the apportioned values ​​of the several leaf nodes included in the first layer.

13. The method according to claim 1, characterized in that The preprocessing includes: mapping processing and normalization processing; and the influence result of the to-be-allocated value of the root node obtained after the normalization processing and the numerator influence result and denominator influence result of each upstream node corresponding to the target node are all in the range of 0 to 1.

14. The method according to any one of claims 1, 11 to 13, characterized in that: The determining the indicator fluctuation analysis result according to the second influence degree includes: The second influence degrees are arranged according to a second preset method, and the indicator fluctuation analysis result is generated according to the arrangement result; wherein the second preset method includes a data flow graph method, a tree graph method or a weighted directed graph method.

15. An index fluctuation analysis device, characterized in that: The index fluctuation analysis device comprises an acquisition unit, a determination unit and an output unit; wherein, The acquisition unit is configured to acquire a large amount of raw data to be analyzed and a corresponding analysis strategy; wherein the analysis strategy at least includes a drill-down analysis strategy and / or an apportionment analysis strategy; The determining unit is configured to determine a first influence of at least one analysis dimension on the overall indicator value when performing indicator fluctuation analysis on the massive raw data using the drill-down analysis strategy, and determine an indicator fluctuation analysis result according to the first influence; The determination unit is further configured to determine a second degree of influence of the allocation result of at least one upstream node on the change of the allocation value of the target node when the allocation analysis strategy is used to perform an indicator fluctuation analysis on the massive raw data, and determine the indicator fluctuation analysis result according to the second degree of influence; the determination unit determines the second degree of influence of the allocation result of at least one upstream node on the change of the allocation value of the target node, including: constructing a data flow network diagram, the data flow network diagram including at least one leaf node; determining a preliminary result of the influence of the change of the allocation value of at least one leaf node; and determining the influence of the change of the allocation value of the at least one leaf node. The preliminary results of the influence degree of the to-be-apported value are subjected to redundant pruning to obtain preliminary results of the influence degree of the root node's to-be-apported value, preliminary results of the numerator influence degree of nodes at each layer, and preliminary results of the denominator influence degree; the preliminary results of the influence degree of the root node's to-be-apported value, preliminary results of the numerator influence degree of nodes at each layer, and preliminary results of the denominator influence degree are preprocessed to obtain the root node's to-be-apported value influence degree result and the numerator influence degree result and denominator influence degree result of each upstream node corresponding to the target node; the root node's to-be-apported value influence degree result and the numerator influence degree result and denominator influence degree result of each upstream node corresponding to the target node are determined as the second influence degree; The output unit is configured to store the indicator fluctuation analysis result and / or display the indicator fluctuation analysis result in a preset manner.

16. An index fluctuation analysis device, characterized in that: The index fluctuation analysis device includes a memory and a processor; wherein, The memory is used to store a computer program that can be run on the processor; The processor is configured to execute the method according to any one of claims 1 to 14 when running the computer program.

17. A computer storage medium, characterized in that: The computer storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 14 is implemented.

18. An index fluctuation analysis system, characterized in that: The indicator fluctuation analysis system at least includes the indicator fluctuation analysis device as described in any one of claims 15 or 16.

Images