An index data self-checking robot system for marketing business early warning monitoring

Abstract

The application relates to the technical field of data analysis, in particular to an index data self-checking robot system for marketing business early warning monitoring, which comprises the following modules: a robot marketing data acquisition module, which acquires various sales data including sales in a preset analysis interval; a robot marketing data classification module, which decomposes a sequence, divides periods according to the fluctuation characteristics of the data, and then classifies; a robot marketing data analysis module, which obtains sales heterogeneity according to the difference characteristics of the data in different categories; compares the results of sequence decomposition to obtain marketing data abnormality; obtains sales data consistency according to the similarity of the marketing data abnormality of different analysis intervals; and a robot marketing data early warning monitoring module, which monitors and warns the state of enterprise marketing business according to the sales data consistency. The application aims to improve the accuracy of robot early warning monitoring of marketing business based on index data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, in particular to a robot system for index data self-checking of marketing business early warning monitoring. BACKGROUND

[0002] Marketing is the front line of company service, and marketing business early warning monitoring is a system for real-time monitoring of marketing activities and related business data, aiming to identify potential problems and abnormal trends in advance to warn the company to take action to avoid or mitigate negative impacts. The monitoring system usually includes continuous tracking of some indicators such as sales, transaction volume, etc., and discovers anomalies in the data by statistical analysis of the indicators.

[0003] In the field of marketing business, existing methods are usually limited to a single data dimension, focusing only on the current marketing business abnormal characteristics, while ignoring the relative characteristics of marketing data and the influence of changes in marketing strategies within a time period. This limitation results in low accuracy of traditional marketing trend analysis in early warning of marketing business problems, and also weakens the reliability of index data self-checking. SUMMARY

[0004] To solve the above technical problems, the present application provides a robot system for index data self-checking of marketing business early warning monitoring to solve the existing problems.

[0005] The present application provides a robot system for index data self-checking of marketing business early warning monitoring, which comprises:

[0006] A robot marketing data collection module for grouping a sequence of various sales data including sales in a preset analysis interval as a sequence of various sales data;

[0007] A robot marketing data classification module for sequence decomposition of the sales data sequence corresponding to the sales, combining the fluctuation characteristics of the sales data, dividing the periods of all sales data sequences; taking all sales data at the same time in each period as a multi-dimensional point, and classifying the multi-dimensional points in each period;

[0008] A robot marketing data analysis module for obtaining sales heterogeneity of each period according to the difference characteristics of sales data in each period of different categories; obtaining marketing data abnormality of each period by combining the results of sequence decomposition of each period and the rest of the periods, and combining the sales heterogeneity; obtaining sales data consistency of each analysis interval according to the similarity of marketing data abnormality of all periods of each analysis interval and the rest of the analysis intervals;

[0009] Robot Marketing Data Early Warning and Monitoring Module: Used to monitor and issue early warnings about the marketing business status of enterprises in the next analysis period based on the consistency of sales data across all analysis periods.

[0010] The specific steps for dividing all sales data sequences into periods include:

[0011] Obtain all peak points in the periodic items of the sequence decomposition result, and take the time period corresponding to two consecutive peak points as a period of various sales data.

[0012] Specifically, the classification of multi-dimensional points within each period is as follows:

[0013] Based on the distance metric between multidimensional points, the multidimensional points in each period are divided into two categories: the category with more multidimensional points is designated as the first category, and the other category is designated as the second category.

[0014] Specifically, the sales heterogeneity obtained for each period is as follows:

[0015] For each period, the data difference coefficient for each period is obtained based on the difference in the number of multi-dimensional points between the two categories;

[0016] Based on the distance distribution between the same sales data of two categories in each period, the purchase difference coefficient for each period is obtained;

[0017] By combining the data difference coefficient and the purchase difference coefficient for each period, the sales heterogeneity for each period can be obtained.

[0018] Specifically, the purchase difference coefficient for each period is obtained as follows:

[0019] The relative differences between various sales data are obtained by measuring the distance between the second category and the first category of sales data; the purchase difference coefficient for each period is obtained by measuring the relative difference of sales revenue for each period and the difference between the remaining relative differences.

[0020] Specifically, the abnormality of marketing data for each period is obtained as follows:

[0021] For each analysis interval, the sales trend similarity coefficient and sales trend difference coefficient for each period are obtained based on the trend term of the sequence decomposition results of each period and the other periods, and the sales heterogeneity.

[0022] The inverse proportional mapping result of the sales trend difference coefficient and the sales trend similarity coefficient for each period is used as the marketing data anomaly degree for each period.

[0023] The acquisition of the sales trend similarity coefficient and sales trend difference coefficient for each period includes:

[0024] For each analysis interval, a sales trend similarity coefficient of each cycle is obtained according to the similarity between the cycle and the corresponding trend item of the sales of all other cycles, and a sales trend difference coefficient of each cycle is obtained according to the difference in sales heterogeneity between the cycle and all other cycles.

[0025] The sales data consistency of the analysis intervals is specifically a reverse proportional mapping result of distance distribution between all marketing data anomaly degrees between the analysis intervals and all other analysis intervals.

[0026] The monitoring and early warning of the business state of the next analysis interval is specifically:

[0027] The sales data consistency of the next analysis interval is predicted according to the distribution of the sales data consistency of all analysis intervals.

[0028] When the normalized result of the sales data consistency of the next analysis interval is less than or equal to a preset abnormal threshold, the robot issues a warning; otherwise, the robot does not issue a warning.

[0029] The prediction of the sales data consistency of the next analysis interval specifically includes the following steps:

[0030] The sales data consistency of all analysis intervals is taken as the input of a sequence prediction algorithm to obtain the sales data consistency of the next analysis interval.

[0031] The application has at least the following beneficial effects:

[0032] The application first obtains multiple sales data in a preset time period and forms a sequence, thereby providing a data basis for subsequent feature analysis of the sales data in different time periods; the sequence is decomposed, which helps to extract more accurate sequence features, divide the data sequence into cycles, measure the periodicity of data changes and fluctuations, and improve the accuracy of subsequent data prediction; all sales data at the same time in each cycle is taken as a multidimensional point, and the relationship and pattern between the data are more intuitively displayed by using visualization technology; the multidimensional points in each cycle are classified to measure the deviation of different data at the same time; the sales heterogeneity is constructed to consider the influence of marketing strategies in different time periods and the correlation and change characteristics of multidimensional marketing data; the marketing data anomaly degree is constructed to analyze the correlation and trend characteristics of the marketing business data in each time period, reflecting the influence of marketing characteristics on marketing business data in different time periods; the sales data consistency is constructed to further consider the regularity characteristics affected by marketing strategies in different time periods, thereby accurately reflecting the abnormal characteristics of marketing business data, accurately predicting the index data of the next time period, and improving the accuracy of the marketing business early warning of the self-checking robot system. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description only constitute some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort.

[0034] Figure 1 A system block diagram of a sales data self-checking robot system for marketing business early warning monitoring is provided for an embodiment of the present application.

[0035] Figure 2 A sales data consistency acquisition process schematic diagram is provided for an embodiment of the present application.

[0036] Figure 3 A specific flowchart of monitoring and early warning of the state of enterprise marketing business is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0037] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purposes, the specific embodiments, structures, features and effects of a sales data self-checking robot system for marketing business early warning monitoring according to the present application are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0039] The specific scheme of a sales data self-checking robot system for marketing business early warning monitoring provided by the present application is described in detail below with reference to the drawings.

[0040] Please refer to Figure 1 which shows a system block diagram of a sales data self-checking robot system for marketing business early warning monitoring provided by an embodiment of the present application. The system comprises a robot marketing data collection module 101, a robot marketing data classification module 102, a robot marketing data analysis module 103, and a robot marketing data early warning monitoring module 104.

[0041] The robot marketing data collection module 101 takes a sequence composed of various sales data including sales in a preset analysis interval as various sales data sequences.

[0042] The present application takes the retail sales data of an enterprise as an example to identify and warn possible abnormal data. It should be noted that the implementer can select other industry and marketing related data for analysis using the same method as the present application.

[0043] The sales data of the enterprise in a preset time period is collected, including sales, product sales quantity and customer flow, and the sales, sales quantity and customer flow of the preset analysis interval are sequentially arranged in ascending order of time to form a sequence, which is denoted as a sales sequence, a sales quantity sequence and a customer flow sequence.

[0044] In some embodiments of the present application, the preset time period is a set of all non-holidays with a time interval between the current time and the preset time period less than or equal to a preset time period, the preset time period is 1 year, and the data collection interval is 1 day; 30 days are taken as an analysis interval; Considering that the sales in the holiday period will increase greatly, which is a predictable and reasonable phenomenon, these peaks and other data are not comparable, so the sales peaks generated by holidays cannot be regarded as abnormal values, so the sales data of each day in the preset time period is collected; The implementer can adjust it by himself.

[0045] Since missing values may occur during the collection process due to external factors, some embodiments of the present application use the isnull() function of the Pandas library to judge the missing values of the above sequences, and if there are missing values in the sequence, the mean filling method is used to fill the missing values. It should be noted that on the basis of realizing the above missing value filling, other ways such as median filling or mode filling can also be used, and the implementer can select them by himself.

[0046] The robot marketing data classification module 102: The sales data sequence corresponding to the sales is decomposed, combined with the fluctuation characteristics of the sales data, and all sales data sequences are divided into periods; All sales data at the same time in each period is taken as a multidimensional point, and the multidimensional points in each period are classified.

[0047] Since sales data is crucial to the strategic decision and resource allocation of an enterprise, it can help the enterprise optimize product portfolio, adjust market strategy and sales channel, and the change of sales can reflect the trend of the market. Sales data may change periodically because consumers have different shopping habits at different times, such as different purchase patterns on weekends and weekdays. Based on this, all sales data sequences are divided into periods according to the sales sequence:

[0048] The sales sequence is decomposed to obtain the trend item and the period item of the sales sequence; All peak points in the period item are obtained, and the time period corresponding to the two consecutive peak points is taken as a period of various sales data.

[0049] It should be noted that there are many methods for sequence decomposition and peak point acquisition, and the implementer can select a suitable method for acquisition, and the present application does not limit this.

[0050] In some embodiments of the present application, a Seasonal and Trend decomposition using Loess (STL) algorithm is used for sequence decomposition. The input of the STL algorithm is the sales sequence of each year, and the output is the trend item, the periodic item and the residual item of the sales sequence. The decomposition obtains each item as a time sequence with the same length as the decomposed sequence. The periodic item shows the periodicity of the change of the sales sequence, and an automatic multiscale-based peak detection (AMPD) algorithm is used to obtain the peak points in the periodic item. It should be noted that the STL sequence decomposition and the AMPD algorithm are known technologies, and the specific process will not be described again.

[0051] Under normal circumstances, various sales data of the marketing business have a certain correlation. For example, when the customer flow is large on a certain day, the possibility of the customer purchase rate increasing is relatively large, and then the sales volume and sales amount will be large, and when the customer flow is small, the sales volume and sales amount will also be small. Due to the fluctuation of the average unit price or the purchase rate of the customers, although different sales data do not show completely strict proportional increase and decrease, the general change trend of different sales data is the same. Based on this, all sales data at the same time in each period is regarded as a multi-dimensional point, and the multi-dimensional points in each period are divided into two categories according to the distance measurement between the multi-dimensional points. The category with a large number of multi-dimensional points is recorded as the first category, and the other category is recorded as the second category.

[0052] In some embodiments of the present application, the sales amount, the sales volume and the customer flow data of the same day are constructed into a three-dimensional point, and all three-dimensional points in a period are mapped to a three-dimensional rectangular coordinate system. Under normal conditions, the distance between the three-dimensional points is relatively close, and the distribution is relatively concentrated. Under abnormal conditions of the marketing business, for example, the customer flow is large on a certain day and the sales amount is small, which causes the three-dimensional point of this day to deviate from other three-dimensional points. Therefore, a fuzzy C-means clustering algorithm of machine learning is used to classify and process the three-dimensional points in each period. The number of clusters is set to 2, that is, the three-dimensional points in each period are divided into two categories. It should be noted that the implementer can also select other classification methods to classify the three-dimensional points, and the present application does not limit this. In addition, the fuzzy C-means clustering algorithm is a known technology, and will not be described again in the present application.

[0053] Further, the category with a large number of three-dimensional points is recorded as the first category, and the category with a small number of three-dimensional points is recorded as the second category.

[0054] The robot marketing data analysis module 103: according to the difference characteristics of the sales data in different categories of each period, the sales heterogeneity of each period is obtained; the results of the sequence decomposition of each period and the remaining periods are integrated, combined with the sales heterogeneity, and the marketing data anomaly degree of each period is obtained; according to the similarity of the marketing data anomaly degree of each analysis interval and all periods of the remaining analysis intervals, the sales data consistency of each analysis interval is obtained.

[0055] For each period, according to the difference between the number of multi-dimensional points of the second category and the first category, the data difference coefficient of each period is obtained; according to the distance measurement between various sales data sequences of the second category and the first category, the relative difference of various sales data is obtained; according to the difference between the relative difference corresponding to the sales of each period and the remaining relative differences, the purchase difference coefficient of each period is obtained; The data difference coefficient and the purchase difference coefficient of each period are fused to obtain the sales heterogeneity of each period.

[0056] It should be noted that the difference represents the numerical difference between variables, including: difference, absolute difference, ratio, etc.; the distance measurement between sequences is a method for measuring the similarity or difference between sequences, including: Euclidean distance, Manhattan distance, DTW distance, etc.; The method of fusion between variables includes: addition, multiplication and mixed addition and multiplication operation; The present application does not limit this.

[0057] In some embodiments of the present application, the ratio of the number of multi-dimensional points of the second category to the first category in each period is taken as the data difference coefficient of each category; The size of the number of multi-dimensional points of the first category reflects how many business days are in the normal state, and the larger the data difference coefficient, the more abnormal days in a period.

[0058] In addition, under normal marketing business, the customer purchase rate is relatively stable, that is, the customer flow is large, and the sales should be large. Therefore, by analyzing the difference between the sales data between the two categories to measure the difference in customer purchase rate stability, and then reflecting the business state in the period. The sequence formed by the sales data in the first category and the second category in ascending order of time is respectively denoted as the sales first sub-sequence and the sales second sub-sequence, and the DTW distance between the sales first sub-sequence and the sales second sub-sequence is taken as the sales relative difference; Correspondingly, the sales volume relative difference and the customer flow relative difference are obtained. The absolute value of the difference between the sales relative difference and the sales volume relative difference, and the customer flow relative difference is calculated, and then averaged to obtain the purchase difference coefficient of each period. The larger the purchase difference coefficient, the more unstable the customer purchase rate corresponding to the first category and the second category in each period.

[0059] Further, the product of the data difference coefficient of each period and the purchase difference coefficient is taken as the sales heterogeneity of each period. The greater the sales heterogeneity, the more unstable the sales state of the business in the period.

[0060] It should be understood that the increase in sales heterogeneity can be caused by changes in the marketing strategy of the enterprise or operational management differences, such as random promotional activities, insufficient supply of goods, and the like, thereby resulting in random rapid growth or decline in sales. If the sales sequence corresponding to the consecutive periods exhibits similar fluctuation characteristics, for example, the marketing strategy of the enterprise causes short-term rapid changes in the data at a specific location in different periods, this characteristic can be abnormal in a period. If the short-term rapid change characteristic occurs regularly, it can be explained as a normal phenomenon. Therefore, the smaller the difference between the sales variation coefficients of different periods, the more regular the short-term rapid change characteristic of sales appears.

[0061] For each analysis interval, the sales trend similarity coefficient of each period is obtained according to the similarity between the trend item corresponding to the sales of each period and the sales of all other periods; the sales trend difference coefficient of each period is obtained according to the difference in sales heterogeneity between each period and all other periods; and the inverse proportion mapping result of the sales trend difference coefficient and the sales trend similarity coefficient of each period is taken as the marketing data abnormality degree of each period.

[0062] It should be noted that the calculation of the similarity between sequences specifically includes the Pearson correlation coefficient, the cosine similarity, the Jaccard similarity, and the like; the inverse proportion mapping result between variables is usually measured by the ratio, the reciprocal, and the like; and the present application does not limit this.

[0063] In some embodiments of the present application, the mean of the absolute values of the cosine similarity of the trend item corresponding to the sales sequence of each period and the sales of all other periods is taken as the sales trend similarity coefficient of each period; and then the mean of the absolute values of the difference in sales heterogeneity between each period and all other periods is taken as the sales trend difference coefficient of each period.

[0064] The ratio of the sales trend difference coefficient and the sales trend similarity coefficient of each period is taken as the marketing data abnormality degree of each period.

[0065] It should be understood that the greater the sales trend similarity coefficient, the more similar the overall trend change of the sales sequence under different periods. The smaller the sales trend difference value, the more likely there are similar variation states of the sales sequence under different periods, and the more regular the short-term rapid change characteristic of sales, and the more normal the marketing business. The greater the sales trend difference value, the more the short-term rapid change characteristic of sales appears only in part of the period, and the more abnormal the marketing business, and the greater the marketing data abnormality degree, indicating that the marketing data is more likely to be abnormal.

[0066] The marketing data anomaly degree focuses on the business data state in the sales cycle, and further, the sales in different analysis intervals have certain similar characteristics.

[0067] The inverse proportion mapping result of the distance distribution of all marketing data anomalies between each analysis interval and all other analysis intervals is taken as the sales data consistency of each analysis interval.

[0068] In some embodiments of the present application, the marketing data anomaly degrees of all cycles in each analysis interval are arranged in ascending order of time to form a marketing data sequence. The marketing business sequence reflects the sales change state characteristics of the enterprise in the analysis interval. Since the consumption habits of users are more similar between different time periods, the inverse of the average Jaccard distance of each analysis interval and the corresponding marketing data sequence of all other analysis intervals in the collected data is taken as the sales data consistency of each analysis interval.

[0069] The acquisition process of the sales data consistency is shown in the schematic diagram as Figure 2 .

[0070] It should be understood that the higher the obtained sales data consistency, the better the overall marketing state of the enterprise in each analysis interval. If the sales data consistency is low, there may be problems in the marketing business of the enterprise in each analysis interval.

[0071] The robot marketing data early warning monitoring module 104 monitors and warns the enterprise marketing business state in the next analysis interval according to the sales data consistency of all analysis intervals.

[0072] The present application analyzes the sales sequence, the sales volume sequence, and the passenger flow sequence of the enterprise, and calculates the sales data consistency in each year. The sales data consistency reflects the possibility of abnormality in the marketing strategy or operation process of the enterprise to a certain extent. The sales data consistency of all analysis intervals is used to predict the sales data consistency of the next analysis interval using the ARIMA model, and normalized using the Z-Score method.

[0073] It should be noted that the implementer can choose other methods to predict the sales data consistency of the next analysis interval, and can also choose other methods to normalize the data, which is not limited by the present application.

[0074] According to the normalized result of the sales data consistency of the next analysis interval, the enterprise marketing business is early warned and monitored;

[0075] If the consistency of the normalized sales data is less than or equal to the preset abnormal threshold, it indicates that the marketing business state is abnormal, and the robot issues a warning to remind the relevant staff of the enterprise to timely adjust the marketing business; if the consistency of the normalized sales data is greater than the preset abnormal threshold, it indicates that the marketing business state is normal, and the robot does not issue a warning.

[0076] In some embodiments of the application, the abnormal threshold value is 0.6; the implementer can adjust it by himself.

[0077] The specific flowchart for monitoring and warning the state of the enterprise marketing business is as shown in Figure 3 .

[0078] To sum up, the embodiments of the application first acquire various sales data in a preset time period and form a sequence; provide a data basis for subsequent feature analysis of the sales data in different time periods; decompose the sequence, which is helpful to extract more accurate sequence features, divide the data sequence into periods, measure the periodicity of the data change fluctuation, and can improve the accuracy of subsequent data prediction; take all sales data at the same time in each period as a multi-dimensional point, use visualization technology to more intuitively show the relationship and pattern between data, classify the multi-dimensional points in each period to measure the deviation of different data at the same time; construct sales heterogeneity, consider the influence of marketing strategies in different time periods and the correlation change characteristics of multi-dimensional marketing data; construct marketing data abnormality, analyze the correlation trend characteristics of the marketing business data of each time period, and reflect the influence of marketing characteristics on marketing business data in different time periods; construct sales data consistency, further consider the law characteristics of the marketing strategy in different time periods, and accurately reflect the abnormal characteristics of the marketing business data, so as to accurately predict the index data of the next time period and improve the accuracy of the marketing business warning of the self-checking robot system.

[0079] It should be noted that the above-mentioned sequence of the embodiments of the application is only for description, and does not represent the advantages and disadvantages of the embodiments. The above describes specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0080] Each embodiment in the present specification is described in a progressive manner, and the same or similar parts of each embodiment can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0081] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; the technical solutions recorded in the foregoing embodiments are modified, or some technical features are replaced equivalently, without making the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and all should be included in the protection scope of the present application.

Claims

1. A system for marketing business early warning monitoring of index data self-inspection robot, characterized in that, The system comprises: a robot marketing data acquisition module: for taking a sequence composed of various sales data including sales in a preset analysis interval as various sales data sequences; a robot marketing data classification module: for performing sequence decomposition on the sales data sequence corresponding to the sales, combining the fluctuation characteristics of the sales data, dividing the periods of all sales data sequences; taking all sales data at the same time in each period as a multi-dimensional point, classifying the multi-dimensional points in each period, dividing the multi-dimensional points in each period into two categories according to the distance measurement between the multi-dimensional points, taking the category with a larger number of multi-dimensional points as the first category, and taking the other category as the second category; a robot marketing data analysis module: for obtaining the sales heterogeneity of each period according to the difference characteristics of the sales data in each period of different categories; obtaining the marketing data abnormality of each period by comprehensively combining the sequence decomposition results of each period and the remaining periods and the sales heterogeneity; obtaining the sales data consistency of each analysis interval according to the similarity of the marketing data abnormality of all periods of each analysis interval and the remaining analysis intervals; a robot marketing data early warning monitoring module: for monitoring and warning the enterprise marketing business state of the next analysis interval according to the sales data consistency of all analysis intervals. The sales heterogeneity of each period is obtained by: for each period, obtaining the data difference coefficient of each period according to the difference in the number of multi-dimensional points of the two categories; obtaining the relative difference of various sales data according to the distance measurement between the various sales data sequences of the second category and the first category; obtaining the purchase difference coefficient of each period according to the difference between the relative difference corresponding to the sales of each period and the remaining relative differences; fusing the data difference coefficient and the purchase difference coefficient of each period to obtain the sales heterogeneity of each period.

2. A machine learning system for marketing business early warning monitoring according to claim 1, wherein, The period of all sales data sequences is divided, and the specific steps include: obtaining all peak points in the period item of the sequence decomposition result, and taking the time period corresponding to the two consecutive peak points as a period of various sales data.

3. A metric data self-checking robot system for marketing business early warning monitoring as claimed in claim 1 wherein, The marketing data abnormality of each period is obtained by: for each analysis interval, obtaining the sales trend similarity coefficient and the sales trend difference coefficient of each period according to the trend item of the sequence decomposition result of each period and the remaining periods and the sales heterogeneity, respectively; taking the inverse proportional mapping result of the sales trend difference coefficient and the sales trend similarity coefficient of each period as the marketing data abnormality of each period.

4. A machine learning system for marketing business early warning monitoring according to claim 3, wherein, The sales trend similarity coefficient and the sales trend difference coefficient of each period include: for each analysis interval, obtaining the sales trend similarity coefficient of each period according to the similarity between the trend items corresponding to the sales of each period and all the remaining periods; obtaining the sales trend difference coefficient of each period according to the difference between the sales heterogeneity of each period and all the remaining periods.

5. A metric data self-checking robot system for marketing business early warning monitoring as claimed in claim 1 wherein, The sales data consistency of each analysis interval is the inverse proportional mapping result of the distance distribution between all marketing data abnormality between each analysis interval and all the remaining analysis intervals.

6. A metric data self-checking robot system for marketing business early warning monitoring as claimed in claim 1 wherein, The monitoring and warning of the enterprise marketing business state of the next analysis interval is specific to: According to the distribution of the sales data consistency of all analysis intervals, the sales data consistency of the next analysis interval is predicted; When the normalized result of the sales data consistency of the next analysis interval is less than or equal to the preset abnormal threshold, the robot issues a warning; otherwise, the robot does not issue a warning.

7. A metric data self-checking robot system for marketing business early warning monitoring as claimed in claim 6 wherein, The prediction of the sales data consistency of the next analysis interval specifically includes the following steps: The sales data consistency of all analysis intervals is taken as the input of a sequence prediction algorithm to obtain the sales data consistency of the next analysis interval.

Images