Customer label generation method based on time sequence
By fitting and adjusting the function model, marking is processed based on customer time factors, the problem of being easily affected by subjective judgment in the existing technology is solved, and the accuracy and success rate of marking is improved.
Patent Information
- Application Number
- CN202410862497.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-28
- Publication Date
- 2024-11-08
AI Technical Summary
The existing marking methods are easily affected by the subjective judgment of the operator, resulting in the customer being mislabeled and reducing the processing success rate.
By obtaining the customer's target processing data, fitting the first function model and the second function model, adjusting the first function model using the second function model, obtaining the third function model, and performing it based on the customer's time factors and the third function model Marking and avoid the influence of subjective judgment.
Improves the accuracy of marking and ensures that customers are accurately labeled, thereby improving processing success rate.
Smart Images

Figure CN118916847A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a method for generating customer tags based on time series. Background Art
[0002] With the development of Internet technology, labeling methods have been widely used. Labeling methods can classify customers into multiple types and handle different types of customers differently.
[0003] In the prior art, labeling methods usually rely on manual labeling by operators, who label customers based on their subjective judgment of the customers.
[0004] However, this method is easily affected by the subjective judgment of operators, resulting in incorrect labeling of customers and incorrect treatment of customers, thereby reducing the success rate of customer treatment. Summary of the invention
[0005] The present application provides a method for generating customer labels based on time series, which solves the problem that the existing labeling method is easily affected by the subjective judgment of the operator, resulting in incorrect labels being placed on the customers.
[0006] In a first aspect, the present application provides a customer label generation method based on time series, comprising: obtaining target processing data corresponding to multiple customers respectively, the target processing data including a time factor affecting target processing of the customer, secondary factors other than the time factor, and a success rate of the target processing of the customer; performing function fitting on the target processing data of multiple customers to obtain a first function model; performing function fitting on the secondary factors and the success rate of multiple customers to obtain a second function model; adjusting the first function model through the second function model to obtain a third function model; and labeling the customer according to the time factor of the customer and the third function model.
[0007] In a possible design, the first function model is adjusted by the second function model to obtain a third function model, including: outputting the first predicted success rate of each of the customers by the first function model; outputting the second predicted success rate of each of the customers by the second function model; performing function fitting by using the first predicted success rates, the time factor and the second predicted success rates corresponding to a plurality of customers respectively to obtain a fourth function model; and adjusting the second function model by the fourth function model to obtain the third function model of the customer.
[0008] In one possible design, the third function model of the customer is obtained by adjusting the second function model through the fourth function model, including: substituting the fourth function model into the second function model to obtain the fifth function model; for each of the customers, substituting the secondary factors of the customer into the fifth function model to obtain the third function model of the customer.
[0009] In a possible design, the customer is marked according to the time factor of the customer and the third function model, including: obtaining at least one critical success rate value; for each of the customers, according to the critical success rate value and the time point corresponding to the peak point of the third function model, dividing the curve corresponding to the third function model of the customer into multiple time periods; the customer is marked according to the time period in which the current time is located, and the marking results corresponding to different time periods are different.
[0010] In one possible design, obtaining at least one critical success rate includes: taking the minimum second predicted success rate among the second predicted success rates of each of the customers as a critical success rate; arranging the second predicted success rates corresponding to each of the customers in ascending order to obtain a success rate sequence; dividing the success rate sequence according to a preset ratio, and taking the third predicted success rate corresponding to the dividing point as another critical success rate.
[0011] In the second aspect, the present application provides a customer label generation device based on time series, including: an acquisition module, used to obtain target processing data corresponding to multiple customers respectively, the target processing data including a time factor affecting the target processing of the customer, secondary factors other than the time factor and a success rate of the target processing of the customer; a first function model fitting module, used to perform function fitting through the target processing data of multiple customers to obtain a first function model; a second function model fitting module, used to perform function fitting through the secondary factors and the success rate of multiple customers to obtain a second function model; a third function model fitting module, used to adjust the first function model through the second function model to obtain a third function model; a labeling module, used to label the customer according to the time factor of the customer and the third function model.
[0012] In a third aspect, the present application provides an electronic device comprising: at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes a time series-based customer tag generation method as described in the first aspect.
[0013] In a fourth aspect, the present application provides a computer storage medium, wherein the computer storage medium stores computer execution instructions. When a processor executes the computer execution instructions, a customer tag generation method based on time series as described in the first aspect is implemented.
[0014] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements a customer tag generation method based on time series as described in the first aspect.
[0015] The present application provides a method for generating customer labels based on time series, which fits the first function model and the second function model through the customer's target processing data, and then adjusts the first function model through the second function model to obtain the third function model, and performs labeling processing on the customer according to the third function model and time factors. The third function model can accurately reflect the relationship between the customer's target processing data and the success rate, and at the same time can avoid the influence of the operator's subjective judgment on the labeling operation, thereby improving the accuracy of the labeling process. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0017] Figure 1 It is a schematic diagram of an application scenario of the method for generating customer tags based on time series provided in an embodiment of the present application;
[0018] Figure 2 A schematic diagram of a method for generating customer tags based on time series provided in one embodiment of the present application;
[0019] Figure 3 A schematic flow chart of a method for generating customer tags based on time series provided in another embodiment of the present application;
[0020] Figure 4 A schematic flow chart of a method for generating customer tags based on time series provided in another embodiment of the present application;
[0021] Figure 5 A schematic diagram of the structure of a time series-based customer tag generation device provided by an embodiment of the present application;
[0022] Figure 6 A schematic diagram of the hardware structure of an electronic device provided for one embodiment of the present application. DETAILED DESCRIPTION
[0023] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0024] The present disclosure is specifically used in a label generation scenario, and different labels can be added to different customers, so as to facilitate different treatments for customers with different labels in the future. In the existing label generation scheme, it is usually based on the operator's subjective judgment of the customer to label the customer. However, this label generation scheme is easily affected by the operator's subjective judgment, resulting in the wrong labeling of the customer, the wrong treatment of the customer, and thus reducing the success rate of the treatment of the customer.
[0025] In order to solve the above technical problems, the embodiment of the present application proposes the following technical solution: a first function model and a second function model are fitted through the target processing data of the customer, and then the first function model is adjusted through the second function model to obtain a third function model, and the customer is marked according to the third function model and the time factor. The third function model can accurately reflect the relationship between the target processing data of the customer and the success rate, and at the same time can avoid the influence of the operator's subjective judgment on the marking operation, thereby improving the accuracy of the marking process.
[0026] Figure 1 Schematic diagram of the application scenario of the method for generating customer tags based on time series provided in the embodiment of the present application. Figure 1 As shown, the server 101 obtains target processing data corresponding to multiple customers respectively; the server 101 performs function fitting through the target processing data of multiple customers to obtain a first function model; the server 101 performs function fitting through the secondary factors and success rates of multiple customers to obtain a second function model; the server 101 adjusts the first function model through the second function model to obtain a third function model; the server 101 performs labeling processing on the customer according to the customer's time factor and the third function model.
[0027] Figure 2 A schematic diagram of a method for generating customer tags based on time series provided in an embodiment of the present application. The execution subject of this embodiment may be Figure 1 The server in the embodiment shown is not particularly limited in this embodiment. Figure 2 As shown, the method includes:
[0028] S201: Obtain target processing data corresponding to a plurality of customers respectively, where the target processing data includes a time factor affecting target processing of customers, secondary factors other than the time factor, and a success rate of target processing of customers.
[0029] Among them, secondary factors refer to all factors that affect target processing except time factors. For example, in the application scenario, the library optimizes services based on customer target processing data, and secondary factors can be the number of books or the damage of books. These secondary factors, like time factors, will affect readers' reading experience.
[0030] Specifically, the target processing data corresponding to the customer can be obtained through direct communication between the operator and the customer. For example, the service personnel of the public transportation system communicate directly with the passengers to obtain the data such as the passenger's travel route, travel time or travel frequency.
[0031] Specifically, the target processing data corresponding to the customer can be obtained by using the digital platform. For example, the passenger's usage records when using the public transportation system, such as the route and frequency of riding, will be recorded in the digital platform of the public transportation system, and the passenger's target processing data can be read from the digital platform.
[0032] S202: Perform function fitting through target processing data of multiple customers to obtain a first function model.
[0033] The first function model is used to indicate the functional relationship between the success rate of target processing for customers and the time factor and the secondary factors.
[0034] Specifically, a spreadsheet software can be used to perform function fitting on the target processing data of multiple customers to obtain a first function model. For example, a function model describing the relationship between the falling speed of an object and time is fitted. The falling speed data is organized into a spreadsheet. It is observed that the data roughly presents a linear relationship, and a linear function is selected as the fitting function. In the spreadsheet software, a scatter plot of the falling speed and time is drawn using the chart function. The spreadsheet software can automatically calculate the parameters of the linear fit and obtain the fitted linear equation, i.e., the first function model.
[0035] Specifically, the database of the programming language can be used to perform function fitting on the target processing data of multiple customers to obtain the first function model. For example, it is necessary to fit a function model of temperature changing over time. The temperature data changes periodically, and the sine function is selected as the fitting function. Using the function in the Python database, a sine function is fitted to the temperature data, and the fitting parameters are optimized by the least squares method to obtain the fitted sine function model, that is, the first function model.
[0036] Specifically, the formula of the first function model may be:
[0037] Y1=K0+K1T+K2X1+K3X2+K4X3+…+K n +X n
[0038] Among them, Y1 is the first prediction success rate output by the first function model, K0, K1, K2, K3, K4, K n is the coefficient, T is the time factor, X1, X2, X3, X n is the secondary factor, and n is the number of secondary factors.
[0039] S203: Perform function fitting through secondary factors and success rates of multiple customers to obtain a second function model.
[0040] The second function model is used to indicate the functional relationship between the success rate and each secondary factor.
[0041] Specifically, linear regression can be used to fit the secondary factors and success rates of multiple customers into a function to obtain a second function model. Linear regression is a method of establishing the relationship between independent variables and dependent variables by fitting a straight line. For example, fitting the relationship between employees' working hours (secondary factors) and work completion (success rate). Linear regression can be used to fit these data points to obtain a linear equation: y=mx+b, where y is work completion, x is working hours, m is the slope, and b is the intercept. Using the least squares method, the values of m and b are calculated to obtain the second function model.
[0042] Specifically, polynomial regression can be used to fit the secondary factors and success rates of multiple customers to obtain a second function model. Polynomial regression is a method of fitting a polynomial curve to establish the relationship between independent variables and dependent variables. For example, fitting the relationship between an athlete's training intensity (secondary factor) and game performance (success rate). There is a nonlinear relationship between training intensity and game performance. You can choose the quadratic polynomial model y=ax 2 +bx+c, where y is the game performance and x is the training intensity. By adjusting the parameters a, b and c, the second function model is obtained.
[0043] Specifically, the formula of the second function model may be:
[0044] Y1'=K0+K1X1+K2X2+K3X3+…+K n X n
[0045] Among them, Y'1 is the second prediction success rate output by the second function model, K0, K1, K2, K3, K n are coefficients, X1, X2, X3, X nis the secondary factor, and n is the number of secondary factors.
[0046] S204: Adjust the first function model through the second function model to obtain a third function model.
[0047] Specifically, when the output of the second function model is the input of the first function model, the second function model can be substituted into the first function model to obtain a third function model. For example, the second function model is y=2x+6, the first function model is z=y+12, the second function model is substituted into the first function model to obtain a third function model z=y+12=2x+6+12=2x+18.
[0048] Specifically, different weights can be assigned to the second function model and the first function model according to their importance, and weighted summation can be performed to obtain the third function model. For example, in employee performance appraisal, the work results calculation formula (second function model) is y=3x+6, and the teamwork performance calculation formula (first function model) is z=8x+9. The weight of work results is 0.7, and the weight of teamwork performance is 0.3. The weighted summation results in employee performance (third function model) of t=0.7y+0.3z=4.5x+6.9.
[0049] S205: Marking the customer according to the customer's time factor and the third function model.
[0050] Among them, labeling means labeling different customers with different labels. For example, labeling employees with good performance as "excellent" and labeling employees with average performance as "ordinary".
[0051] Specifically, customers can be labeled based on the cluster analysis model. For example, the customer's time factor data can be substituted into the cluster analysis model, and the customers can be divided into different groups through the calculation of the model, and corresponding labels can be generated for each customer group. For example, the reader's borrowing time can be substituted into the cluster model, and the readers can be divided into several different groups through the calculation of the model, and each group represents a borrowing time pattern. According to the clustering results, each group is labeled, such as "morning borrowing group", "evening active group" and other labels.
[0052] Specifically, historical data can be used to train a neural network model (third function model). In this model, the input is time and the output is the corresponding label. For example, there is a set of historical temperature data, including different time points and corresponding temperature labels, which can be used to train a neural network. After the model training is completed, a time point is input to the model, such as 1 o'clock, and the model will output the corresponding temperature label, such as the label of "low temperature"; if 12 o'clock is input, the model will output the label of "high temperature".
[0053] In summary, by fitting the first function model and the second function model using the target processing data of multiple customers, and then using the second function model to accurately adjust the first function model, the third function model finally obtained can more comprehensively consider the various factors affecting the success rate. Based on this more accurate model, the customer's marking process can improve the accuracy of marking.
[0054] Figure 3 A schematic flow chart of a method for generating customer tags based on time series provided in another embodiment of the present application. Figure 2 Based on the provided embodiment, this embodiment focuses on describing the specific process of adjusting the first function model through the second function model to obtain the third function model, including:
[0055] S301: Outputting a first predicted success rate for each customer through a first function model.
[0056] Specifically, the user's target processing data can be input into the first function model, and the output of the first function model is the first prediction success rate. For example, the library predicts the success rate of readers renewing books. The target processing data such as readers' borrowing history, reading habits, and book type preferences are input into the regression model, and the regression model can output the success rate of readers renewing books.
[0057] Specifically, the customer's target processing data may include multiple data. Multiple data are input into the first function model to obtain multiple output values. The average of the multiple output values can be taken as the first prediction success rate. For example, to predict the success rate of employee promotion, the employee's target processing data includes task completion and teamwork. The employee's task completion is input into the first function model, and the output value is 70%; the employee's teamwork is input into the first function model, and the output value is 80%. Calculate the average value and get the employee's promotion success rate of 75%.
[0058] S302: Outputting a second predicted success rate for each customer through a second function model.
[0059] Specifically, the user's target processing data can be input into the second function model, and the output of the second function model is the second prediction success rate. For example, the success rate of the athlete's performance improvement is predicted. The target processing data such as the athlete's historical performance, technical level, training time, etc. are input into the regression model, and the regression model can output the success rate of the athlete's performance improvement.
[0060] Specifically, the customer's target processing data can include multiple data. Multiple data are input into the second function model to obtain multiple output values. The average of the multiple output values can be taken as the second prediction success rate. For example, to predict the success rate of increasing product output, the target processing data includes the use of new technologies and the work of employees. The use of new technologies is input into the second function model, and the output value is 50%; the work of employees is input into the second function model, and the output value is 60%. Calculate the average value and get a success rate of increasing product output of 55%.
[0061] S303: Perform function fitting based on the first prediction success rates, time factors, and second prediction success rates corresponding to a plurality of customers to obtain a fourth function model.
[0062] Specifically, the first prediction success rate, time factor, and second prediction success rate of each customer can be used as input variables, and a target variable related to customer performance (such as customer satisfaction, business growth rate, etc.) can be used as an output variable to perform polynomial fitting to obtain the fourth function model. For example, the job seeker's academic qualification matching success rate, job browsing time, and work experience matching success rate can be used as inputs, and the interview success rate can be used as output to perform polynomial fitting to obtain the fourth function model.
[0063] Specifically, a neural network model can be constructed, taking the first prediction success rate, time factor and second prediction success rate as input, and training the network to learn the relationship between these features and the target variable. For example, predicting a customer's shopping success rate. The first prediction success rate is calculated based on the customer's historical shopping records, and the second prediction success rate is calculated based on the customer's current browsing time and browsing times. The first prediction success rate, time factor and second prediction success rate are taken as input features, and the relationship between these features and the customer's shopping success rate is learned by training the network to obtain the fourth function model.
[0064] Specifically, the formula of the fourth function model obtained by performing function fitting based on the first prediction success rate Y1, the time factor T and the second prediction success rate Y'1 corresponding to a plurality of customers may be:
[0065] Y1=Y1'+L1T+L2T 2 +L3T 3 …+L n T n
[0066] Among them, Y1 is the first prediction success rate, Y'1 is the second prediction success rate, L1, L2, L3 are coefficients, T is the time factor, and n is the number of secondary factors.
[0067] S304: The second function model is adjusted by the fourth function model to obtain the third function model of the customer.
[0068] Specifically, when the output of the fourth function model is the input of the second function model, the fourth function model can be substituted into the second function model to obtain the third function model. For example, the fourth function model is y=2x, the second function model is z=y+10, and the fourth function model is substituted into the second function model to obtain the third function model z=y+10=2x+10.
[0069] Specifically, the customer's third function model is obtained by adjusting the second function model through the fourth function model, which is described in detail as follows: Sa~Sb:
[0070] Sa: Substitute the fourth function model into the second function model to obtain the fifth function model.
[0071] Specifically, the fourth function model can be directly substituted into the second function model to obtain the fifth function model. For example, the second function model is y=t+2x, and the fourth function model is t=3x+2. The fourth function model is directly substituted into the second function model to obtain the fifth function model y=5x+2.
[0072] Specifically, a weight may be set for the fourth function model, and the fourth function model is multiplied by the weight and inserted into the second function model to obtain the fifth function model. For example, a weight of 0.5 is set for the fourth function model y=2x, and the fourth function model y=2x is multiplied by 0.5 and inserted into the third function model t=2y to obtain the fifth function model t=2x.
[0073] Specifically, the formula of the fifth function model can be:
[0074] Y1=K0+K1X1+K2X2+K3X3+…+K n X n +L1T+L2T 2 +L3T 3 …+L n T n
[0075] Among them, Y1 is the first prediction success rate, X1, X2, X3, X n As a secondary factor, K0, K1, K2, K3, K n is the coefficient of the secondary factor, T is the time factor, L1, L2, L3 are the coefficients of the time factor T, and n is the number of secondary factors.
[0076] Sb: For each customer, substitute the customer's secondary factors into the fifth function model to obtain the customer's third function model.
[0077] Specifically, the secondary factors of each customer can be directly substituted into the fifth function model to obtain the third function model. For example, the fifth function model is t=ax+by, where a and b are secondary factors, and x and y are independent variables. The secondary factors of customer A are a=2 and b=3. Substituting these secondary factors into the fifth function model, the third function model of customer A is obtained as t=2x+3y.
[0078] Specifically, different weights can be set for the various secondary factors according to the importance of the various secondary factors of the customer, and the weighted sum of the various secondary factors can be inserted into the fifth function model to obtain the third function model of the customer. For example, the fifth function model is a model that predicts a person's health based on his or her daily habits. Now, adjust this model according to secondary factors such as exercise frequency and eating habits to obtain the third function model. Eating habits are assigned a weight of 0.5; exercise frequency is assigned a weight of 0.3. Multiply the specific score of each secondary factor by its corresponding weight, and then add all the products to obtain a weighted sum. Substitute this weighted sum into the fifth function model to obtain the third function model.
[0079] In summary, the first function model and the second function model consider different influencing factors (time factors and other minor factors) to predict the success rate of customers. By combining these two models and adjusting them using the fourth function model, a more comprehensive and accurate third function model can be obtained. Such a multi-model combination can provide more comprehensive and rich information, which helps to improve the accuracy of success rate calculation.
[0080] Figure 4 A schematic flow chart of a method for generating customer tags based on time series provided in another embodiment of the present application. Figure 2 Based on the provided embodiment, this embodiment focuses on describing the specific process of marking the customer according to the customer's time factor and the third function model, including:
[0081] S401: Obtain at least one critical success rate value.
[0082] Specifically, the critical value that affects the success rate can be determined by collecting and analyzing historical data. For example, if a library wants to improve readers’ borrowing satisfaction, it can be found through analysis that when the library has 100,000 books, the borrowing satisfaction rate is significantly improved. Then, these 100,000 books can be regarded as a critical value for the success rate.
[0083] Specifically, at least one critical success rate value is obtained, which is described in detail as follows: Sc~Se:
[0084] Sc: The minimum second prediction success rate among the second prediction success rates of each customer is taken as a success rate critical value.
[0085] Specifically, the minimum second prediction success rate among the second prediction success rates can be found out as a success rate critical value through the judgment of the operator. For example, the second prediction success rates include: 5%, 10%, 15%, 20%. Through the judgment of the operator, it can be concluded that the success rate critical value is 5%.
[0086] Specifically, the second prediction success rate can be input into a computer system, and the minimum second prediction success rate among the second prediction success rates can be found through the minimum value function of the computer system as a critical value of the success rate. For example, the second prediction success rates include: 9%, 90%, 85%, and 26%. The minimum second prediction success rate of 9% among the second prediction success rates can be found through the minimum value function of the computer system, and the critical value of the success rate can be obtained as 9%.
[0087] Sd: Arrange the second predicted success rates corresponding to each customer in ascending order to obtain a success rate sequence.
[0088] Specifically, the second prediction success rates corresponding to the customers can be arranged from small to large by the judgment of the operator to obtain a success rate sequence. For example, the second prediction success rates are: 95%, 18%, 55%, 2%. The operator arranges the second prediction success rates from small to large to obtain a success rate sequence of: 2%, 18%, 55%, 95%.
[0089] Specifically, the second prediction success rates corresponding to the customers can be arranged from small to large by using the ascending function of the computer system to obtain a success rate sequence. For example, the second prediction success rates are: 90%, 17%, 45%, 29%. The second prediction success rates are arranged from small to large by using the ascending function of the computer system to obtain a success rate sequence of: 17%, 29%, 45%, 90%.
[0090] Se: Split the success rate sequence according to a preset ratio, and use the third predicted success rate corresponding to the split point as another success rate critical value.
[0091] Specifically, the total number of success rate sequences can be determined. According to the preset ratio, the position of the split point is directly calculated. After finding the split point position, check the third predicted success rate corresponding to the position and use it as the new success rate critical value. For example, there is a third predicted success rate sequence containing 100 points, the total number of points is 100, and the split ratios are 25%, 50%, 75% and 100%. The split point positions are calculated as: the 25th point (25%*100), the 50th point (50%*100), the 75th point (75%*100) and the 100th point (100%*100). In the third predicted success rate sequence, find the success rate values of the 25th point, the 50th point, the 75th point and the 100th point, which are the success rate critical values.
[0092] Specifically, a graphical auxiliary tool can be used to segment the success rate sequence, and the third predicted success rate corresponding to the segmentation point is used as another critical value of the success rate. For example, the success rate sequence is plotted as a line graph or a scatter plot, and the segmentation point position corresponding to the preset ratio is manually marked. If the preset ratio is 60%, then find the position of about 60% on the graph. The third predicted success rate at the segmentation point is the critical value of the success rate.
[0093] S402: For each customer, according to the critical value of the success rate and the time point corresponding to the peak point of the third function model, divide the curve corresponding to the third function model of the customer into multiple time periods.
[0094] Specifically, the curve corresponding to the third function model can be divided into multiple time periods according to multiple time points corresponding to multiple success rate critical points and the time point corresponding to the peak point of the third function model. For example, the curve corresponding to the third function model is a parabola, the time points corresponding to the success rate critical points are 3 and 9, and the time point corresponding to the peak point of the third function model is 6. Through these three time points, the parabola corresponding to the third function model can be divided into 4 time periods, namely: (-∞, 3], (3, 6], (6, 9] and (9, +∞).
[0095] Specifically, the segmentation of the curve of the third function model can be realized based on the success rate critical value, peak point and the increase and decrease of the curve. Determine the increase and decrease of the curve corresponding to the third function model before and after the success rate critical value and peak point, that is, whether the model is rising or falling to divide the time period. For example, the values of the third function model within a period of time are as follows:
[0096] T: 1, 2, 3, 4, 5, 6.
[0097] S: 50, 55, 65, 60, 55, 50.
[0098] Where T is time and S is success rate. The critical value of success rate is 60 and the peak value is 65. Time period 1: T = 1 to T = 2, the success rate is in the rising stage but the success rate is lower than the critical value; Time period 2: T = 3, the success rate reaches the peak point and the success rate is higher than the critical value; Time period 3: T = 4 to T = 6, the success rate begins to decline and the success rate is lower than the critical value.
[0099] S403: Marking is performed on the customer according to the time period of the current time, and different time periods correspond to different marking results.
[0100] Specifically, you can define time period rules, such as "early morning", "morning", "afternoon", "evening", etc., and set the corresponding time range for each rule. Get the current time. Compare the current time with the defined time period rules to determine the time period to which it belongs. Label the customer according to the time period. For example, the following time period rules are defined: early morning: 00:00-06:00, morning: 06:00-12:00, afternoon: 12:00-18:00, evening: 18:00-24:00. The current time is 14:30, which belongs to the "afternoon" time period. Therefore, the customer is marked as an "afternoon customer".
[0101] Specifically, you can define personalized time period rules based on business needs or customer characteristics. These rules may be dynamically adjusted based on the customer's active time, shopping habits, etc. Get the current time. Compare the current time with the customer's personalized time period rules to determine the time period it belongs to. Label the customer according to the time period. For example, based on the customer's shopping history analysis, a customer is usually active in the following time periods: morning shopping peak: 08:00-10:00, lunch break shopping period: 13:00-14:30, evening leisure shopping: 20:00-22:00. The current time is 15:45, which does not belong to the customer's active time period according to the customer's personalized time period rules. Therefore, you can choose not to label or label it with a "non-active time period". If the current time is 21:00, it belongs to the customer's "evening leisure shopping" time period, and the customer is marked as "evening leisure shopping customer".
[0102] In summary, by introducing the success rate critical value, the differences in customer success rates in different time periods can be more accurately defined. Based on these differences, the marking process can more accurately reflect the characteristics of customers in different time periods, improving the accuracy of marking.
[0103] Figure 5 This is a schematic diagram of the structure of a time series-based customer tag generation device provided by an embodiment of the present application. Figure 5 As shown, the time series-based customer tag generation device includes: an acquisition module 501 , a first function model fitting module 502 , a second function model fitting module 503 , a third function model fitting module 504 and a labeling module 505 .
[0104] The acquisition module 501 is used to acquire target processing data corresponding to multiple customers respectively, and the target processing data includes time factors affecting target processing of customers, secondary factors other than time factors, and success rate of target processing of customers.
[0105] The first function model fitting module 502 is used to perform function fitting through target processing data of multiple customers to obtain a first function model.
[0106] The second function model fitting module 503 is used to perform function fitting through the secondary factors and success rates of multiple customers to obtain a second function model.
[0107] The third function model fitting module 504 is used to adjust the first function model through the second function model to obtain a third function model.
[0108] The marking module 505 is used to mark the customer according to the customer's time factor and the third function model.
[0109] In one embodiment, the third function model fitting module 504 includes: a first success rate prediction unit, used to output the first predicted success rate of each customer through the first function model; a second success rate prediction unit, used to output the second predicted success rate of each customer through the second function model; a fourth function model fitting unit, used to perform function fitting through the first predicted success rates, time factors and second predicted success rates corresponding to multiple customers respectively to obtain a fourth function model; a third function model calculation unit, used to adjust the second function model through the fourth function model to obtain the third function model of the customer.
[0110] In one embodiment, the third function model calculation unit includes: a fifth function model calculation component, used to substitute the fourth function model into the second function model to obtain the fifth function model; and a third function model calculation component, used to substitute the customer's secondary factors into the fifth function model for each customer, to obtain the customer's third function model.
[0111] In one embodiment, the marking module 505 includes: a success rate critical value acquisition unit, used to obtain at least one success rate critical value; a curve division unit, used to divide the curve corresponding to the customer's third function model into multiple time periods for each customer according to the success rate critical value and the time point corresponding to the peak point of the third function model; a marking unit, used to mark the customer according to the time period in which the current time is located, and the marking results corresponding to different time periods are different.
[0112] In one embodiment, a success rate critical value acquisition unit includes: a first success rate critical value calculation component, used to take the minimum second predicted success rate among the second predicted success rates of each customer as a success rate critical value; a sorting component, used to arrange the second predicted success rates corresponding to each customer in ascending order to obtain a success rate sequence; a second success rate critical value calculation component, used to divide the success rate sequence according to a preset ratio, and take the third predicted success rate corresponding to the dividing point as another success rate critical value.
[0113] The device provided in this embodiment can be used to execute the technical solution of the above method embodiment. Its implementation principle and technical effect are similar, and this embodiment will not be repeated here.
[0114] Figure 6 The hardware structure diagram of an electronic device provided in one embodiment of the present application is shown in FIG. Figure 6 As shown, the electronic device of this embodiment includes: at least one processor 601 and a memory 602; the memory stores computer execution instructions; at least one processor executes the computer execution instructions stored in the memory, so that at least one processor executes the above time series-based customer label generation method.
[0115] Optionally, the memory 602 may be independent or integrated with the processor 601 .
[0116] When the memory 602 is independently provided, the service device further includes a bus 603 for connecting the memory 602 and the processor 601 .
[0117] The embodiment of the present application further provides a computer storage medium, in which computer execution instructions are stored. When a processor executes the computer execution instructions, the above-mentioned method for generating customer tags based on time series is implemented.
[0118] An embodiment of the present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the method for generating customer tags based on time series as described above is implemented.
[0119] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.
[0120] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to implement the solution of this embodiment.
[0121] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The above-mentioned module-composed unit can be implemented in the form of hardware or in the form of hardware plus software functional units.
[0122] The above-mentioned integrated module implemented in the form of a software function module can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium, including a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform some steps of the methods of various embodiments of the present application.
[0123] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be implemented by a combination of hardware and software modules in the processor.
[0124] The memory may include a high-speed RAM memory, and may also include a non-volatile storage NVM, such as at least one disk memory, and may also be a USB flash drive, a mobile hard disk, a read-only memory, a magnetic disk or an optical disk, etc.
[0125] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.
[0126] The above storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The storage medium can be any available medium that can be accessed by a general or special purpose computer.
[0127] An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the storage medium can also exist as discrete components in an electronic device or a main control device.
[0128] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.
[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.
Claims
1. A customer tag generation method based on time series, characterized in that: include: Acquire target processing data corresponding to a plurality of customers respectively, wherein the target processing data includes a time factor affecting target processing of the customers, secondary factors other than the time factor, and a success rate of the target processing of the customers; Performing function fitting on the target processed data of the plurality of customers to obtain a first function model; Performing function fitting by using the secondary factors and the success rates of the plurality of customers to obtain a second function model; Adjusting the first function model by using the second function model to obtain a third function model; The customer is marked according to the time factor of the customer and the third function model.
2. The method according to claim 1, characterized in that The step of adjusting the first function model by using the second function model to obtain a third function model includes: Outputting a first predicted success rate for each of the customers through the first function model; Outputting a second predicted success rate for each of the customers through the second function model; A fourth function model is obtained by performing function fitting based on the first prediction success rates, the time factor, and the second prediction success rates respectively corresponding to the plurality of customers; The second function model is adjusted by the fourth function model to obtain the third function model of the customer.
3. The method according to claim 2, characterized in that The step of adjusting the second function model by using the fourth function model to obtain the third function model of the customer includes: Substituting the fourth function model into the second function model to obtain a fifth function model; For each of the customers, the secondary factors of the customer are substituted into the fifth function model to obtain the third function model of the customer.
4. The method according to claim 2 or 3, characterized in that: The step of marking the customer according to the time factor of the customer and the third function model includes: obtaining at least one critical success rate value; For each of the customers, according to the critical value of the success rate and the time point corresponding to the peak point of the third function model, the curve corresponding to the third function model of the customer is divided into a plurality of time periods; The customer is marked according to the time period of the current time, and different time periods correspond to different marking results.
5. The method according to claim 4, characterized in that The obtaining of at least one critical success rate value comprises: taking the minimum second prediction success rate among the second prediction success rates of the customers as a success rate critical value; Arrange the second predicted success rates corresponding to the respective customers in ascending order to obtain a success rate sequence; The success rate sequence is divided according to a preset ratio, and the third predicted success rate corresponding to the division point is used as another success rate critical value.
6. A customer tag generation device based on time series, characterized in that: include: An acquisition module, used for acquiring target processing data corresponding to a plurality of customers respectively, wherein the target processing data includes a time factor affecting the target processing of the customer, a secondary factor other than the time factor, and a success rate of the target processing of the customer; A first function model fitting module, used for performing function fitting on the target processed data of the plurality of customers to obtain a first function model; A second function model fitting module, used for performing function fitting through the secondary factors and the success rates of the plurality of customers to obtain a second function model; A third function model fitting module, used for adjusting the first function model by using the second function model to obtain a third function model; The marking module is used to mark the customer according to the time factor of the customer and the third function model.
7. An electronic device, characterized in that: include: at least one processor and memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method for generating customer tags based on time series as described in any one of claims 1 to 5.
8. A computer storage medium, characterized in that: The computer storage medium stores computer executable instructions. When the processor executes the computer executable instructions, the method for generating customer tags based on time series as described in any one of claims 1 to 5 is implemented.
9. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements a method for generating customer tags based on time series as claimed in any one of claims 1 to 5.
