Gift exchange early warning method and device, storage medium and computer equipment
By establishing a point redemption prediction model based on the autoregressive sliding average model, the problems of excessive gift redemption cycle and gift reserve management in financial institutions are solved, effective management of gift reserves is achieved, and redemption efficiency and user experience are improved.
Patent Information
- Application Number
- CN202410884830.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-03
- Publication Date
- 2024-11-08
AI Technical Summary
In financial institutions, too long gift redemption cycle or too much gift reserve will affect user experience and cause property losses, and it is difficult for existing technology to effectively predict and manage gift redemption needs.
By obtaining the user's historical points redemption data, establish a point redemption prediction model based on the autoregressive sliding average model, predict future points, and send warning information based on the difference between the expected number of redeemed gifts and the existing number of gifts, and adjust gift reserves in advance.
Effectively estimate whether the gift reserve is sufficient, avoid the problem of insufficient gift distribution, improve redemption efficiency, improve user experience and reduce property losses.
Smart Images

Figure CN118917886A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of financial technology, and in particular to a gift redemption early warning method and device, storage medium, and computer equipment. Background Art
[0002] In today's banks and other financial institutions, users usually purchase a variety of financial products from financial institutions. Different financial products can be exchanged for corresponding user points, and users can exchange points for other gifts, such as USB flash drives, small ornaments, etc. Since the purchase cycle for exchanging gifts can sometimes be long, if users wait too long for the gift distribution cycle, it will directly affect the user's overall experience of the financial institution. At the same time, as more and more users purchase financial products, more and more users use points to exchange gifts. If gifts are not reserved in advance, the gift exchange cycle will be long and the gifts cannot be distributed in time, affecting the user's experience in the point exchange process. If too many gifts are reserved and some gifts have time limits, it will cause financial institutions to suffer property losses. Summary of the invention
[0003] In view of this, the present application provides a gift exchange warning method and device, storage medium, and computer equipment to obtain the user's historical points exchange data; sum the exchange points within the same preset points integration period to obtain the historical points exchange integration data, and establish a user points exchange prediction model based on the historical points exchange integration data after data cleaning and the autoregressive sliding average model; based on the user points exchange prediction model, predict the user's expected exchange points in the future preset period, and when the difference between the number of gifts to be exchanged corresponding to the expected exchange points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, send a gift distribution warning message to the preset recipient terminal. By introducing the autoregressive sliding average model to predict the expected exchange points, and then obtain the number of gifts to be exchanged, it is possible to estimate in advance whether the number of retained gifts is sufficient, avoid the problem of insufficient gift distribution, and improve the exchange efficiency.
[0004] According to one aspect of the present application, a gift exchange early warning method is provided, the method comprising:
[0005] Obtain the user's historical points redemption data, wherein the historical points redemption data includes redemption points and redemption time;
[0006] Sum up the redemption points redeemed in the same preset point integration period, and obtain historical point redemption integration data based on the summed total redemption points of each of the multiple preset point integration periods, wherein the total redemption points of each preset point integration period in the historical point redemption integration data are arranged in chronological order, and the time span of each preset point integration period is the same;
[0007] Performing data cleaning on the historical points exchange integration data, and establishing a user points exchange prediction model based on the historical points exchange integration data after data cleaning and an autoregressive moving average model;
[0008] Based on the user points redemption prediction model, the user's expected redemption points in the future preset period are predicted. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal.
[0009] According to another aspect of the present application, a gift exchange warning device is provided, the device comprising:
[0010] The points data acquisition module is used to acquire the user's historical points redemption data, wherein the historical points redemption data includes the redemption points and the redemption time;
[0011] A points data integration module, used to sum up the redemption points redeemed within the same preset points integration period, and obtain historical points redemption integration data based on the summed total points redeemed in each of the multiple preset points integration periods, wherein the total points redeemed in each preset points integration period in the historical points redemption integration data are arranged in chronological order, and the time span of each preset points integration period is the same;
[0012] A prediction model building module is used to clean the historical points exchange integrated data, and to build a user points exchange prediction model based on the historical points exchange integrated data after data cleaning and an autoregressive sliding average model;
[0013] The gift redemption warning module is used to predict the user's expected redemption points in a future preset period based on the user points redemption prediction model. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal.
[0014] According to another aspect of the present application, a storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the above-mentioned gift redemption warning method is implemented.
[0015] According to another aspect of the present application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the above-mentioned gift redemption warning method when executing the program.
[0016] By means of the above technical scheme, the present application provides a gift exchange warning method and device, storage medium, and computer equipment to obtain the user's historical points exchange data; sum the exchange points within the same preset points integration period to obtain the historical points exchange integration data, and establish a user points exchange prediction model based on the historical points exchange integration data after data cleaning and the autoregressive sliding average model; based on the user points exchange prediction model, predict the user's expected exchange points in the future preset period, and when the difference between the number of gifts to be exchanged corresponding to the expected exchange points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, send a gift distribution warning message to the preset recipient terminal. By introducing the autoregressive sliding average model to predict the expected exchange points, and then obtain the number of gifts to be exchanged, it is possible to estimate in advance whether the number of retained gifts is sufficient, avoid the problem of insufficient gift distribution, and improve the exchange efficiency.
[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0019] Figure 1 A flowchart of a gift exchange warning method provided in an embodiment of the present application is shown;
[0020] Figure 2 A schematic diagram of a flow chart of another gift exchange warning method provided in an embodiment of the present application is shown;
[0021] Figure 3 A flow chart of another gift exchange warning method provided in an embodiment of the present application is shown;
[0022] Figure 4 A flow chart of another gift exchange warning method provided in an embodiment of the present application is shown;
[0023] Figure 5 A schematic diagram of the structure of a gift exchange warning device provided in an embodiment of the present application is shown;
[0024] Figure 6 A structural schematic diagram of another gift redemption warning device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION
[0025] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.
[0026] In this embodiment, a gift exchange early warning method is provided, such as Figure 1 As shown, the method includes:
[0027] Step 101, obtaining the user's historical points redemption data, wherein the historical points redemption data includes redemption points and redemption time.
[0028] In the above embodiments of the present application, it can be applied to the gift distribution system of financial institutions, such as banks, etc. In order to maintain the stability and efficiency of the gift distribution system, it is necessary to ensure that there is an appropriate gift inventory, so that even if the demand for gift exchange suddenly increases, the financial institution can smoothly redeem the gift distribution and improve the user experience.
[0029] Specifically, the user's historical points redemption data is obtained, where the historical points redemption data includes redemption points and redemption time. For example, the points redemption situation of existing users of a financial institution in the past five years can be obtained, including the redemption amount and redemption time of each point, to prepare for the establishment of a subsequent user points redemption prediction model.
[0030] Step 102, sum up the redemption points within the same preset point integration period, and obtain historical point redemption integration data based on the sum of the total redemption points of multiple preset point integration periods, wherein the total redemption points of each preset point integration period in the historical point redemption integration data are arranged in chronological order, and the time span of each preset point integration period is the same.
[0031] Next, the acquired data is integrated to further reduce the amount of original data and improve computing efficiency. Specifically, the redemption points with redemption time in the same preset points integration period are summed to obtain the total redemption points for the preset points integration period, and the historical points redemption integration data is obtained based on the total redemption points of each of the multiple preset points integration periods. The time span of each preset points integration period can be set to 5 minutes, or other time spans can be set according to the needs of the actual application scenario. For example, for each preset points integration period from 1:05 to 1:55 on January 1, 2022, and the total redemption points of each preset points integration period, as shown in Table 1:
[0032] Table 1
[0033] Preset points integration period Total points redeemed 2022-1-1 1:05-1:10 1289 2022-1-1 1:10-1:15 1249 2022-1-1 1:15-1:20 1259 2022-1-1 1:20-1:25 1649 2022-1-1 1:25-1:30 1749 2022-1-1 1:30-1:35 1849 2022-1-1 1:35-1:40 1949 2022-1-1 1:40-1:45 1309 2022-1-1 1:45-1:50 1319 2022-1-1 1:50-1:55 1389
[0034] At the same time, when integrating data, time is a key dimension. The time sequence is used as the key (Key), and the total redemption points corresponding to each time point is used as the value (Value) to organize the historical points redemption integration data, that is, sorting according to the time sequence of each preset points integration period, so that the data can be organized and accessed in chronological order, which facilitates subsequent analysis.
[0035] Step 103, clean the historical points exchange integrated data, and establish a user points exchange prediction model based on the historical points exchange integrated data after data cleaning and the autoregressive moving average model.
[0036] Next, the historical points redemption integrated data is cleaned, such as removing invalid or abnormal data in the historical points redemption integrated data, and processing missing data, so as to improve the quality and completeness of the data. Based on the historical points redemption integrated data after data cleaning and the autoregressive moving average model, a user points redemption prediction model is established.
[0037] The Auto-Regressive Moving Average Mode (ARMA) is a time series model that combines the auto-regressive model AR (Auto-Regressive Mode) and the moving average model MA (Moving Average Mode). The auto-regressive model is used to describe the correlation between the current value and the previous value. It can express the current value as a linear combination of the previous values and consider the influence of the error term. It can capture the autocorrelation in the time series, that is, the dependence between the current value and the past value. The moving average model is used to describe the correlation between the current value and the current error term value and the previous error term value. By expressing the current value as a linear combination of the current error term and the past error term, it can capture the lagged error in the time series, which can help understand the relationship between the current value and the past error term, as well as the influence of the past error term on the current value. The auto-regressive moving average model combines the advantages of the auto-regressive model and the moving average model, and can capture the autocorrelation and lagged error in the time series at the same time.
[0038] By using historical points redemption integration data, the autoregressive moving average model can analyze the changes in users' future points redemption, that is, by modeling and analyzing historical data, it can predict the trend of future data changes. Since the autoregressive moving average model can capture the autocorrelation and lag errors in the time series, it has a stronger ability to predict data changes. Compared with the traditional time series model, the autoregressive moving average model can more accurately describe the changing patterns of data and provide more reliable prediction results.
[0039] Step 104, based on the user points redemption prediction model, predict the user's expected redemption points in a future preset period. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, send a gift distribution warning message to the preset recipient terminal.
[0040] Next, based on the user points redemption prediction model, predict the user's expected redemption points within the future preset period. The prediction time can be set to every preset period. The preset period can be set to 1 hour, 1 day or 2 days, or other time periods set according to the actual application scenario, and can be changed at any time. For example, when the preset period is set to 1 day, the user's expected redemption points on the next day are predicted every day.
[0041] When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal. For this reason, if a large number of users redeem points and a large number of redemption gifts are required, if it is judged that the current number of retained gifts is insufficient to meet the required number of redemption gifts, the gift distribution personnel are required to purchase or allocate the gifts to be redeemed in advance. By timely predicting the number of gifts to be redeemed corresponding to the expected redemption points, the redemption gifts can be managed more effectively, which can improve the user experience and the efficiency of gift distribution, and reduce property losses.
[0042] By applying the technical solution of this embodiment, the user's historical points redemption data is obtained; the redemption points within the same preset points integration period are summed to obtain the historical points redemption integration data, and a user points redemption prediction model is established based on the historical points redemption integration data after data cleaning and the autoregressive sliding average model; based on the user points redemption prediction model, the user's expected redemption points in the future preset period are predicted, and when the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal. By introducing the autoregressive sliding average model to predict the expected redemption points and then obtain the number of gifts to be redeemed, it is possible to estimate in advance whether the number of retained gifts is sufficient, avoid the problem of insufficient gift distribution, and improve redemption efficiency.
[0043] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another gift exchange warning method is provided, such as Figure 2 As shown, the method includes:
[0044] Step 201, obtaining the user's historical points redemption data, wherein the historical points redemption data includes redemption points and redemption time.
[0045] Step 202, sum up the redemption points within the same preset point integration period, and obtain historical point redemption integration data based on the sum of the total redemption points of multiple preset point integration periods, wherein the total redemption points of each preset point integration period in the historical point redemption integration data are arranged in chronological order, and the time span of each preset point integration period is the same.
[0046] In the above embodiment of the present application, the historical points redemption data of the user is obtained, wherein the historical points redemption data includes the redemption points and the redemption time, and the redemption points of the redemption time in the same preset points integration period are summed to obtain the total redemption points of the preset points integration period, and the historical points redemption integration data is obtained based on the total redemption points of each of the multiple preset points integration periods, so as to prepare for the establishment of the subsequent user points redemption prediction model.
[0047] Step 203, clean the historical points exchange integrated data to obtain multiple sets of candidate order arrays, and for any candidate order array, calculate the Akaike information value corresponding to the candidate order array based on the historical points exchange integrated data after data cleaning, the autoregressive moving average model, the candidate order array and the Akaike information calculation formula.
[0048] Step 204: for each of the Akaike information values corresponding to each group of candidate order arrays, the candidate order array corresponding to the minimum Akaike information value is determined as the optimal order array of the autoregressive moving average model, wherein the Akaike information calculation formula is:
[0049] AIC(i,j)=2k-2ln(L),k=i+j+1,
[0050] AIC(i,j) is the Akaike information value corresponding to the alternative order array (i,j), i and j are both natural numbers, k is the parameter in the autoregressive moving average model, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical integral exchange integrated data after data cleaning and the alternative order array (i,j).
[0051] Next, the historical points exchange integration data is cleaned to obtain multiple sets of candidate order arrays, such as (1, 2), (2, 1), (1, 0), etc. For any candidate order array, the Akaike information value corresponding to the candidate order array is calculated based on the historical points exchange integration data after data cleaning, the autoregressive moving average model, the candidate order array and the Akaike information calculation formula, and the candidate order array corresponding to the minimum Akaike information value is determined as the optimal order array of the autoregressive moving average model. The Akaike Information Criterion (AIC) is suitable for application scenarios with more complex models and has higher calculation accuracy.
[0052] Step 205: Based on the autoregressive moving average model of the optimal order array, a user points redemption prediction model is established. Based on the user points redemption prediction model, the user's expected redemption points in a future preset period are predicted, wherein the autoregressive moving average model includes an autoregressive model and a moving average model, and the autoregressive moving average model is:
[0053] Y t,ARMA(p,q) =c+φ1y t-1 +φ2y t-2 +...+φ p y t-p +ε t +θ1y t ' -1 +θ2y t ' -2 +...+θ q y t ' -q ,
[0054] Y t,ARMA(p,q) Represents the time series of the autoregressive moving average model with the optimal order array (p,q), where p and q are the autoregressive order and the moving average order respectively, c is the preset constant term, φ1, φ2...φ p are the coefficients of the 1st to the pth autoregressive terms, y t-1 ,y t-2 ...y t-p are the autoregressive terms from t-1th to tpth, θ1, θ2...θ q are the coefficients of the 1st to qth moving average terms, y t ' -1 ,y t ' -2 ...y t ' -p are the moving average terms from t-1th to tqth, ε t is the preset noise term, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
[0055] Next, based on the autoregressive moving average model with the optimal order array, a user points redemption prediction model is established, and the user's expected points to be redeemed in the future preset period are predicted based on the user points redemption prediction model. In particular, the preset noise term of the autoregressive moving average model is obtained by superimposing and interfering the preset noise terms of the autoregressive model AR and the moving average model MA.
[0056] Step 206: When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, the number of gifts to be allocated is obtained based on the difference between the number of gifts to be redeemed and the number of gifts retained in the current preset period, and gift distribution warning information is generated based on the type of gifts to be redeemed, the number of gifts to be allocated and the predicted future preset period, and the gift distribution warning information is sent to the preset recipient terminal.
[0057] Finally, when the difference between the number of gifts to be redeemed corresponding to the expected redemption points (800) and the number of gifts retained in the current preset period (200) is greater than or equal to the preset risk value (for example, 500), the number of gifts to be allocated (600) is obtained based on the difference (600) between the number of gifts to be redeemed (800) and the number of gifts retained in the current preset period (200), and gift distribution warning information is generated based on the type of gifts to be redeemed (U disk, calendar, etc., based on the current gift distribution settings set by the financial institution), the number of gifts to be allocated and the predicted future preset period, and the gift distribution warning information is sent to the preset recipient terminal. To this end, by setting a preset risk value, that is, the number of gifts that the financial institution can currently prepare immediately, and then based on the established user points redemption prediction model, the expected redemption points in the future are estimated according to the different dimensions of the preset period. For example, the possible changes in points redemption in the next hour, the next whole day or the next two days are predicted respectively, and the number of gifts to be redeemed is obtained and compared with the current number of gifts retained. If the difference exceeds the set preset risk value, it means that the financial institution cannot quickly prepare the expected number of gifts for distribution, and an alarm mechanism is immediately activated to remind the gift distribution management personnel that gifts need to be allocated so that the gifts can be distributed in time.
[0058] By applying the technical solution of this embodiment, the historical points redemption data of the user is obtained, the redemption points within the same preset points integration period are summed up, the historical points redemption integration data is obtained, and multiple sets of candidate order arrays are obtained. For any candidate order array, the Akaike information value corresponding to the candidate order array is calculated based on the historical points redemption integration data after data cleaning, the autoregressive moving average model, the candidate order array and the Akaike information calculation formula, and the candidate order array corresponding to the minimum Akaike information value is determined as the optimal order array of the autoregressive moving average model. Based on the autoregressive moving average model that determines the optimal order array, a user points redemption prediction model is established. Based on the user points redemption prediction model, it is predicted that the user will have a high probability of redeeming in the future. The expected redemption points within the preset period. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, the number of gifts to be allocated is obtained based on the difference between the number of gifts to be redeemed and the number of gifts retained in the current preset period. Gift distribution warning information is generated based on the type of gifts to be redeemed, the number of gifts to be allocated and the predicted future preset period, and the gift distribution warning information is sent to the preset recipient terminal. By introducing the autoregressive sliding average model to establish a user points redemption prediction model, and performing time series prediction on the number of gifts to be redeemed corresponding to the expected redemption points in the future, it is possible to better monitor whether the current retained gifts are sufficient.
[0059] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another gift exchange warning method is provided, such as Figure 3 As shown, the method includes:
[0060] Step 301, obtaining the user's historical points redemption data, wherein the historical points redemption data includes redemption points and redemption time.
[0061] Step 302, sum up the redemption points within the same preset point integration period, and obtain historical point redemption integration data based on the sum of the total redemption points of multiple preset point integration periods, wherein the total redemption points of each preset point integration period in the historical point redemption integration data are arranged in chronological order, and the time span of each preset point integration period is the same.
[0062] In the above embodiment of the present application, the user's historical points redemption data is obtained, the redemption points within the same preset points integration period are summed up, and the total redemption points for the preset points integration period are obtained. Based on the total redemption points of each of the multiple preset points integration periods, the historical points redemption integration data is obtained to prepare for the follow-up.
[0063] Step 303, calculating the second-order difference of the total points exchanged in each preset points integration period in the historical points exchange integration data based on the second-order difference calculation formula.
[0064] Step 304: When the second-order difference of the total points exchanged in any preset points integration period is greater than the second-order difference mean, the total points exchanged in the preset points integration period greater than the second-order difference mean are eliminated to obtain the historical points exchange integration data after data cleaning, wherein the second-order difference calculation formula is:
[0065] Δx f 2 =Δx f -Δx f-1 ,f∈(1,2,...,t),
[0066] Δx f 2 is the second-order difference of the total points redeemed during the f-th preset points integration period, Δx f and Δx f-1 are the total points redeemed during the f-th and f-1-th preset points integration periods respectively, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
[0067] Next, the historical points exchange integration data is standardized to ensure that the different features or variables in the historical points exchange integration data are comparable, thereby laying a solid foundation for subsequent analysis. In order to eliminate the trend term and cycle term in the historical points exchange integration data, the first-order or second-order difference technique can be used. The first-order difference refers to the difference between the values of adjacent time points, while the second-order difference is the difference of the first-order difference again. Both methods help stabilize the data and make it more suitable for time series analysis. Specifically, based on the second-order difference calculation formula, the second-order difference of the total points exchanged in each preset points integration period in the historical points exchange integration data is calculated. When the second-order difference of the total points exchanged in any preset points integration period is greater than the mean of the second-order difference, the total points exchanged in the preset points integration period greater than the mean of the second-order difference are eliminated to obtain the historical points exchange integration data after data cleaning, which is used to eliminate the unsmoothed data. After the unsmoothed data is eliminated, the subsequent prediction accuracy can be improved. In particular, after the difference processing is performed, the stability of the historical points exchange integration data after data cleaning can also be tested. Stationarity is a basic assumption in time series analysis, which means that the statistical properties of the data (such as mean and variance) do not change over time. Through the stationarity test, we can verify whether the data meets this condition.
[0068] Step 305, obtain multiple sets of candidate order arrays, and for any candidate order array, calculate the Bayesian information value corresponding to the candidate order array based on the historical points exchange integrated data after data cleaning, the autoregressive moving average model, the candidate order array and the Bayesian information calculation formula.
[0069] Step 306, for each of the Bayesian information values corresponding to each group of candidate order arrays, the candidate order array corresponding to the minimum Bayesian information value is determined as the optimal order array of the autoregressive moving average model, wherein the Bayesian information calculation formula is:
[0070] BIC(m,l)=k ln(t)-2ln(L),k=m+l+1,
[0071] BIC(m,l) is the Akaike information value corresponding to the alternative order array (m,l), m and l are both natural numbers, k is the parameter in the autoregressive moving average model, t is the total number of preset point integration periods in the historical point exchange integration data after data cleaning, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical point exchange integration data after data cleaning and the alternative order array (m,l).
[0072] Next, multiple sets of candidate order arrays are obtained. For any candidate order array, the Bayesian information value corresponding to the candidate order array is calculated based on the historical points exchange integration data after data cleaning, the autoregressive moving average model, the candidate order array and the Bayesian information calculation formula, and the candidate order array corresponding to the minimum Bayesian information value is determined as the optimal order array of the autoregressive moving average model. The Bayesian Information Criterion (BIC) is fast in operation and is suitable for application scenarios where the prediction accuracy is not particularly high but the prediction results need to be obtained quickly.
[0073] Step 307: Establish a user points redemption prediction model based on the autoregressive moving average model of the optimal order array, wherein the autoregressive moving average model is:
[0074] Y t,ARMA(p,q) =c+φ1y t-1 +φ2y t-2 +...+φ p y t-p +ε t +θ1y t ' -1 +θ2y t ' -2 +...+θ q y t ' -q ,
[0075] Yt,ARMA(p,q) Represents the time series of the autoregressive moving average model with the optimal order array (p,q), where p and q are the autoregressive order and the moving average order respectively, c is the preset constant term, φ1, φ2...φ p are the coefficients of the 1st to the pth autoregressive terms, y t-1 ,y t-2 ...y t-p are the autoregressive terms from t-1th to tpth, θ1, θ2...θ q are the coefficients of the 1st to qth moving average terms, y t ' -1 ,y t ' -2 ...y t ' -p are the moving average terms from t-1th to tqth, ε t is the preset noise term, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
[0076] Next, based on the autoregressive moving average model that determines the optimal order array, a user points redemption prediction model is established. The purpose of determining the order is to estimate the p and q values. The order with the smallest value can be determined by methods such as the AIC criterion or the BIC criterion, and the data order is the data order. In particular, when using the historical points redemption integrated data after data cleaning to establish a user points redemption prediction model, in order to ensure that the entire data processing process is dynamic and can continuously adapt to changes over time, it is also possible to periodically reacquire "new" historical points redemption integrated data, such as updating the historical points redemption integrated data every preset period (half a year, a year or other preset period), and re-establishing the user points redemption prediction model, so that the user points redemption prediction model can be "updated in real time" and then predict data that is more in line with reality.
[0077] Step 308, based on the user points redemption prediction model, predict the user's expected redemption points in the future preset period. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, send gift distribution warning information to the preset recipient terminal.
[0078] Next, based on the user points redemption prediction model, the user's expected redemption points in the future preset period are predicted. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning information is sent to the preset recipient terminal.
[0079] By applying the technical solution of this embodiment, the historical points redemption data of the user is obtained. The redemption points with the same redemption time in the same preset points integration period are summed to obtain the historical points redemption integration data, and the second-order difference of the total points redeemed in each preset points integration period in the historical points redemption integration data is calculated based on the second-order difference calculation formula, and the total points redeemed in the preset points integration period greater than the second-order difference mean are eliminated to obtain the historical points redemption integration data after data cleaning, and multiple groups of candidate order arrays are obtained. For any candidate order array, the Bayesian information value corresponding to the candidate order array is calculated based on the historical points redemption integration data after data cleaning, the autoregressive sliding average model, the candidate order array and the Bayesian information calculation formula. The alternative order array corresponding to the minimum Bayesian information value is determined as the optimal order array of the autoregressive moving average model. Based on the autoregressive moving average model that determines the optimal order array, a user points redemption prediction model is established to predict the user's expected redemption points in a preset period in the future, and a gift distribution warning information is sent to the preset recipient terminal. By introducing the autoregressive moving average model to establish a user points redemption prediction model, and performing time series prediction on the number of gifts to be redeemed corresponding to the future expected redemption points, it is possible to better monitor whether the current retained redemption gifts are sufficient.
[0080] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another gift exchange warning method is provided, such as Figure 4 As shown, the method includes:
[0081] Step 401, obtaining the user's historical points redemption data, wherein the historical points redemption data includes redemption points and redemption time.
[0082] Step 402, summing up the redemption points within the same preset point integration period, and obtaining historical point redemption integration data based on the summed total redemption points of multiple preset point integration periods, wherein the total redemption points of each preset point integration period in the historical point redemption integration data are arranged in chronological order, and the time span of each preset point integration period is the same.
[0083] In the above embodiment of the present application, the user's historical points redemption data is obtained, the redemption points within the same preset points integration period are summed up, and the total redemption points for the preset points integration period are obtained. Based on the total redemption points of each of the multiple preset points integration periods, the historical points redemption integration data is obtained to prepare for the follow-up.
[0084] Step 403, clean the historical points exchange integrated data, determine the data autocorrelation characteristics of the historical points exchange integrated data after data cleaning based on the historical points exchange integrated data after data cleaning and the autocorrelation function, and determine the data partial autocorrelation characteristics of the historical points exchange integrated data after data cleaning based on the historical points exchange integrated data after data cleaning and the partial autocorrelation function.
[0085] Step 404, when the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the existence of data truncation characteristics, and the data partial autocorrelation characteristics show the existence of data tailing characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the autoregressive model.
[0086] Step 405, when the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the existence of data tailing characteristics, and the data partial autocorrelation characteristics show the existence of data truncation characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the moving average model.
[0087] Step 406, when the data autocorrelation characteristics and the partial autocorrelation characteristics of the cleaned historical point redemption integrated data both show the existence of data tailing characteristics, a user point redemption prediction model is established based on the cleaned historical point redemption integrated data and the autoregressive moving average model, wherein the data autocorrelation characteristics and the partial autocorrelation characteristics of the cleaned historical point redemption integrated data do not simultaneously show data truncation characteristics.
[0088] Next, when selecting the order of the ARMA model, it is usually necessary to observe the sample autocorrelation function (ACF) and partial autocorrelation function (PACF). When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the existence of data truncation characteristics, and the data partial autocorrelation characteristics show the existence of data tailing characteristics, the autoregressive model is used to establish a user points exchange prediction model. When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the existence of data tailing characteristics, and the data partial autocorrelation characteristics show the existence of data truncation characteristics, the moving average model is used to establish a user points exchange prediction model. When the data autocorrelation characteristics and data partial autocorrelation characteristics of the cleaned historical points exchange integrated data both show the existence of data tailing characteristics, the autoregressive sliding average model is used to establish a user points exchange prediction model, which can determine the best model for different application scenarios. At the same time, when the model is established, other methods such as cross-validation can also be used to assist model selection.
[0089] Step 407, based on the user points redemption prediction model, predict the user's expected redemption points in the future preset period. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, send gift distribution warning information to the preset recipient terminal.
[0090] Finally, based on the user points redemption prediction model, the user's expected redemption points in the future preset period are predicted. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning information is sent to the preset recipient terminal.
[0091] By applying the technical solution of this embodiment, the historical points redemption data of the user is obtained, and the redemption points with redemption time within the same preset points integration period are summed to obtain the historical points redemption integration data. Based on the historical points redemption integration data after data cleaning and the autocorrelation function, the data autocorrelation characteristics of the historical points redemption integration data after data cleaning are determined, and based on the historical points redemption integration data after data cleaning and the partial autocorrelation function, the data partial autocorrelation characteristics of the historical points redemption integration data after data cleaning are determined. When the data autocorrelation characteristics of the cleaned historical points redemption integration data show the existence of data truncation characteristics, and the data partial autocorrelation characteristics show the existence of data tailing characteristics, a user points redemption prediction model is established based on the autoregressive model. When the data autocorrelation characteristics of the cleaned historical points redemption integration data show the existence of data tailing characteristics, and the data partial autocorrelation characteristics show the existence of data truncation characteristics, a user points redemption prediction model is established based on the moving average model. When the data autocorrelation characteristics and partial autocorrelation characteristics of the cleaned historical points redemption integrated data both show the existence of data tailing characteristics, a user points redemption prediction model is established based on the autoregressive sliding average model. The user's expected redemption points within a preset period in the future are predicted based on the user points redemption prediction model, and gift distribution warning information is sent to the preset recipient terminal. By introducing the autoregressive sliding average model, a user points redemption prediction model is established to improve the efficiency of gift distribution.
[0092] Further, as Figure 1 The specific implementation of the method, the embodiment of the present application provides a gift exchange warning device, such as Figure 5 As shown, the device comprises:
[0093] Points data acquisition module 501, used to acquire historical points redemption data of the user, wherein the historical points redemption data includes redemption points and redemption time;
[0094] The points data integration module 502 is used to sum the redemption points redeemed within the same preset points integration period, and obtain historical points redemption integration data based on the summed total points redeemed in multiple preset points integration periods, wherein the total points redeemed in each preset points integration period in the historical points redemption integration data are arranged in chronological order, and the time span of each preset points integration period is the same;
[0095] The prediction model building module 503 is used to clean the historical points exchange integrated data, and build a user points exchange prediction model based on the historical points exchange integrated data after data cleaning and the autoregressive moving average model;
[0096] The gift redemption warning module 504 is used to predict the user's expected redemption points in a future preset period based on the user points redemption prediction model. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal.
[0097] Optionally, the prediction model building module 503 is further used to:
[0098] Based on the historical points exchange integration data after data cleaning and the autocorrelation function, determine the data autocorrelation characteristics of the historical points exchange integration data after data cleaning, and based on the historical points exchange integration data after data cleaning and the partial autocorrelation function, determine the data partial autocorrelation characteristics of the historical points exchange integration data after data cleaning;
[0099] When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the presence of data truncation characteristics, and the data partial autocorrelation characteristics show the presence of data tailing characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the autoregressive model;
[0100] When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the presence of data tailing characteristics, and the data partial autocorrelation characteristics show the presence of data truncation characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the moving average model;
[0101] When the data autocorrelation characteristics and partial autocorrelation characteristics of the historical point redemption integrated data after data cleaning both show the existence of data tailing characteristics, a user point redemption prediction model is established based on the historical point redemption integrated data after data cleaning and the autoregressive moving average model, wherein the data autocorrelation characteristics and partial autocorrelation characteristics of the historical point redemption integrated data after data cleaning do not simultaneously show data truncation characteristics.
[0102] Optionally, the prediction model building module 503 is further used to:
[0103] Based on the historical points exchange integration data after data cleaning and the autoregressive moving average model, determine the optimal order array of the autoregressive moving average model;
[0104] Based on the autoregressive moving average model of the optimal order array, a user points redemption prediction model is established, wherein the autoregressive moving average model is:
[0105] Y t,ARMA(p,q) =c+φ1y t-1 +φ2y t-2 +...+φ p y t-p +ε t +θ1y t ' -1 +θ2y t ' -2 +...+θ q y t ' -q ,
[0106] Y t,ARMA(p,q) Represents the time series output term of the autoregressive moving average model with the optimal order array (p,q), where p and q are the autoregressive order and the moving average order respectively, c is the preset constant term, φ1, φ2...φ p are the coefficients of the 1st to the pth autoregressive terms, y t-1 ,y t-2 ...y t-p are the autoregressive terms from t-1th to tpth, θ1, θ2...θ q are the coefficients of the 1st to qth moving average terms, y t ' -1 ,y t ' -2 ...y t ' -p are the moving average terms from t-1th to tqth, ε t is the preset noise term, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
[0107] Optionally, the prediction model building module 503 is further used to:
[0108] Acquire multiple sets of candidate order arrays, and for any candidate order array, calculate the Akaike information value corresponding to the candidate order array based on the historical points exchange integrated data after data cleaning, the autoregressive moving average model, the candidate order array and the Akaike information calculation formula;
[0109] For each group of candidate order arrays corresponding to the Akaike information value, the candidate order array corresponding to the minimum Akaike information value is determined as the optimal order array of the autoregressive moving average model, wherein the Akaike information calculation formula is:
[0110] AIC(i,j)=2k-2ln(L),k=i+j+1,
[0111] AIC(i,j) is the Akaike information value corresponding to the alternative order array (i,j), i and j are both natural numbers, k is the parameter in the autoregressive moving average model, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical integral exchange integrated data after data cleaning and the alternative order array (i,j).
[0112] Optionally, the prediction model building module 503 is further used to:
[0113] Obtain multiple sets of candidate order arrays, and for any candidate order array, calculate the Bayesian information value corresponding to the candidate order array based on the historical points exchange integrated data after data cleaning, the autoregressive moving average model, the candidate order array and the Bayesian information calculation formula;
[0114] For each group of candidate order arrays corresponding to the Bayesian information value, the candidate order array corresponding to the minimum Bayesian information value is determined as the optimal order array of the autoregressive moving average model, wherein the Bayesian information calculation formula is:
[0115] BIC(m,l)=k ln(t)-2ln(L),k=m+l+1,
[0116] BIC(m,l) is the Akaike information value corresponding to the alternative order array (m,l), m and l are both natural numbers, k is the parameter in the autoregressive moving average model, t is the total number of preset point integration periods in the historical point exchange integration data after data cleaning, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical point exchange integration data after data cleaning and the alternative order array (m,l).
[0117] Optionally, the gift warning module 504 is further used to:
[0118] When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, the number of gifts to be allocated is obtained based on the difference between the number of gifts to be redeemed and the number of gifts retained in the current preset period, and gift distribution warning information is generated based on the type of gifts to be redeemed, the number of gifts to be allocated and the predicted future preset period, and the gift distribution warning information is sent to the preset recipient terminal.
[0119] Furthermore, the present application embodiment provides another gift exchange warning device, such as Figure 6 As shown, the device comprises:
[0120] The points data acquisition module 601 is used to acquire the historical points redemption data of the user, wherein the historical points redemption data includes the redemption points and the redemption time;
[0121] The points data integration module 602 is used to sum the redemption points redeemed within the same preset points integration period, and obtain historical points redemption integration data based on the summed total points redeemed in each of the multiple preset points integration periods, wherein the total points redeemed in each preset points integration period in the historical points redemption integration data are arranged in chronological order, and the time span of each preset points integration period is the same;
[0122] The prediction model building module 603 is used to clean the historical points exchange integrated data, and build a user points exchange prediction model based on the historical points exchange integrated data after data cleaning and the autoregressive moving average model;
[0123] The gift redemption warning module 604 is used to predict the user's expected redemption points in a future preset period based on the user points redemption prediction model. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal.
[0124] The points data cleaning module 605 is used to calculate the second-order difference of the total points exchanged in each preset points integration period in the historical points exchange integration data based on the second-order difference calculation formula. When the second-order difference of the total points exchanged in any preset points integration period is greater than the second-order difference mean, the total points exchanged in the preset points integration period greater than the second-order difference mean are eliminated to obtain the historical points exchange integration data after data cleaning, wherein the second-order difference calculation formula is:
[0125] Δx f 2 =Δx f -Δx f-1 ,f∈(1,2,...,t),
[0126] Δx f 2 is the second-order difference of the total points redeemed during the f-th preset points integration period, Δx f and Δx f-1 are the total points redeemed during the f-th and f-1-th preset points integration periods respectively, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
[0127] Optionally, the prediction model building module 603 is further used for:
[0128] Based on the historical points exchange integration data after data cleaning and the autocorrelation function, determine the data autocorrelation characteristics of the historical points exchange integration data after data cleaning, and based on the historical points exchange integration data after data cleaning and the partial autocorrelation function, determine the data partial autocorrelation characteristics of the historical points exchange integration data after data cleaning;
[0129] When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the presence of data truncation characteristics, and the data partial autocorrelation characteristics show the presence of data tailing characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the autoregressive model;
[0130] When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the presence of data tailing characteristics, and the data partial autocorrelation characteristics show the presence of data truncation characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the moving average model;
[0131] When the data autocorrelation characteristics and partial autocorrelation characteristics of the historical point redemption integrated data after data cleaning both show the existence of data tailing characteristics, a user point redemption prediction model is established based on the historical point redemption integrated data after data cleaning and the autoregressive moving average model, wherein the data autocorrelation characteristics and partial autocorrelation characteristics of the historical point redemption integrated data after data cleaning do not simultaneously show data truncation characteristics.
[0132] Optionally, the prediction model building module 603 is further used to:
[0133] Based on the historical points exchange integration data after data cleaning and the autoregressive moving average model, determine the optimal order array of the autoregressive moving average model;
[0134] Based on the autoregressive moving average model of the optimal order array, a user points redemption prediction model is established, wherein the autoregressive moving average model is:
[0135] Y t,ARMA(p,q) =c+φ1y t-1 +φ2y t-2 +...+φ p y t-p +ε t +θ1y t ' -1 +θ2y t ' -2 +...+θ q y t ' -q ,
[0136] Y t,ARMA(p,q) Represents the time series output term of the autoregressive moving average model with the optimal order array (p,q), where p and q are the autoregressive order and the moving average order respectively, c is the preset constant term, φ1, φ2...φ p are the coefficients of the 1st to the pth autoregressive terms, y t-1 ,y t-2 ...y t-p are the autoregressive terms from t-1th to tpth, θ1, θ2...θ q are the coefficients of the 1st to qth moving average terms, y t ' -1 ,y t ' -2 ...y t ' -p are the moving average terms from t-1th to tqth, ε t is the preset noise term, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
[0137] Optionally, the prediction model building module 603 is further used to:
[0138] Acquire multiple sets of candidate order arrays, and for any candidate order array, calculate the Akaike information value corresponding to the candidate order array based on the historical points exchange integrated data after data cleaning, the autoregressive moving average model, the candidate order array and the Akaike information calculation formula;
[0139] For each group of candidate order arrays corresponding to the Akaike information value, the candidate order array corresponding to the minimum Akaike information value is determined as the optimal order array of the autoregressive moving average model, wherein the Akaike information calculation formula is:
[0140] AIC(i,j)=2k-2ln(L),k=i+j+1,
[0141] AIC(i,j) is the Akaike information value corresponding to the alternative order array (i,j), i and j are both natural numbers, k is the parameter in the autoregressive moving average model, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical integral exchange integrated data after data cleaning and the alternative order array (i,j).
[0142] Optionally, the prediction model building module 603 is further used to:
[0143] Obtain multiple sets of candidate order arrays, and for any candidate order array, calculate the Bayesian information value corresponding to the candidate order array based on the historical points exchange integrated data after data cleaning, the autoregressive moving average model, the candidate order array and the Bayesian information calculation formula;
[0144] For each group of candidate order arrays corresponding to the Bayesian information value, the candidate order array corresponding to the minimum Bayesian information value is determined as the optimal order array of the autoregressive moving average model, wherein the Bayesian information calculation formula is:
[0145] BIC(m,l)=k ln(t)-2ln(L),k=m+l+1,
[0146] BIC(m,l) is the Akaike information value corresponding to the alternative order array (m,l), m and l are both natural numbers, k is the parameter in the autoregressive moving average model, t is the total number of preset point integration periods in the historical point exchange integration data after data cleaning, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical point exchange integration data after data cleaning and the alternative order array (m,l).
[0147] Optionally, the gift warning module 604 is further used to:
[0148] When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, the number of gifts to be allocated is obtained based on the difference between the number of gifts to be redeemed and the number of gifts retained in the current preset period, and gift distribution warning information is generated based on the type of gifts to be redeemed, the number of gifts to be allocated and the predicted future preset period, and the gift distribution warning information is sent to the preset recipient terminal.
[0149] It should be noted that for other corresponding descriptions of the functional units involved in the gift exchange warning device provided in the embodiment of the present application, please refer to Figures 1 to 4 The corresponding description in the method will not be repeated here.
[0150] Based on the above Figures 1 to 4 The method shown in the embodiment of the present application is accordingly provided with a storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned Figures 1 to 4 The gift redemption warning method shown.
[0151] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.
[0152] Based on the above Figures 1 to 4 The method shown, and Figure 5 and Figure 6In order to achieve the above-mentioned purpose, the embodiment of the present application further provides a computer device, which can be a personal computer, a server, a network device, etc. The computer device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figures 1 to 4 The gift redemption warning method shown.
[0153] Optionally, the computer device may further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a WI-FI interface), etc.
[0154] Those skilled in the art will appreciate that the computer device structure provided in this embodiment does not limit the computer device, and may include more or fewer components, or a combination of certain components, or different component arrangements.
[0155] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and saves the hardware and software resources of the computer device, and supports the operation of information processing programs and other software and / or programs. The network communication module is used to realize communication between the components inside the storage medium, and communication with other hardware and software in the physical device.
[0156] Through the description of the above implementation methods, the technical personnel in this field can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or can be implemented by hardware to obtain the user's historical points redemption data; sum the redemption points within the same preset points integration period, and obtain the historical points redemption integration data, and establish a user points redemption prediction model based on the historical points redemption integration data after data cleaning and the autoregressive sliding average model; based on the user points redemption prediction model, predict the user's expected redemption points in the future preset period, and when the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, send a gift distribution warning message to the preset recipient terminal. By introducing the autoregressive sliding average model to predict the expected redemption points, and then obtaining the number of gifts to be redeemed, it is possible to estimate in advance whether the number of retained gifts is sufficient, avoid the problem of insufficient gift distribution, and improve the redemption efficiency.
[0157] Those skilled in the art will appreciate that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily necessary for implementing the present application. Those skilled in the art will appreciate that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the description of the implementation scenario, or can be changed accordingly and located in one or more devices different from the present implementation scenario. The modules of the above-mentioned implementation scenario can be integrated into one module, or can be further split into multiple submodules.
[0158] The above serial numbers of this application are only for description and do not represent the advantages and disadvantages of the implementation scenarios. The above disclosure is only a few specific implementation scenarios of this application, but this application is not limited to them, and any changes that can be thought of by technicians in this field should fall within the scope of protection of this application.
Claims
1. A gift exchange warning method, It is characterized in that The method comprises: Obtain the user's historical points redemption data, wherein the historical points redemption data includes redemption points and redemption time; Sum up the redemption points redeemed in the same preset point integration period, and obtain historical point redemption integration data based on the summed total redemption points of each of the multiple preset point integration periods, wherein the total redemption points of each preset point integration period in the historical point redemption integration data are arranged in chronological order, and the time span of each preset point integration period is the same; Performing data cleaning on the historical points exchange integration data, and establishing a user points exchange prediction model based on the historical points exchange integration data after data cleaning and an autoregressive moving average model; Based on the user points redemption prediction model, the user's expected redemption points in the future preset period are predicted. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal.
2. The method according to claim 1, It is characterized in that The data cleaning of the historical points redemption integrated data includes: Calculate the second-order difference of the total points exchanged in each preset points integration period in the historical points exchange integration data based on the second-order difference calculation formula; When the second-order difference of the total points exchanged in any preset points integration period is greater than the second-order difference mean, the total points exchanged in the preset points integration period greater than the second-order difference mean are eliminated to obtain the historical points exchange integration data after data cleaning, wherein the second-order difference calculation formula is: Δx f 2 =Δx f -Δx f-1 ,f∈(1,2,...,t), Δx f 2 is the second-order difference of the total points redeemed during the f-th preset points integration period, Δx f and Δx f-1 are the total points redeemed during the f-th and f-1-th preset points integration periods respectively, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
3. The method according to claim 1, It is characterized in that Before establishing a user points exchange prediction model based on the historical points exchange integration data after data cleaning and the autoregressive sliding average model, the method further includes: Based on the historical points exchange integration data after data cleaning and the autocorrelation function, determine the data autocorrelation characteristics of the historical points exchange integration data after data cleaning, and based on the historical points exchange integration data after data cleaning and the partial autocorrelation function, determine the data partial autocorrelation characteristics of the historical points exchange integration data after data cleaning; Accordingly, the user points exchange prediction model is established based on the historical points exchange integration data after data cleaning and the autoregressive moving average model, including: When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the presence of data truncation characteristics, and the data partial autocorrelation characteristics show the presence of data tailing characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the autoregressive model; When the data autocorrelation characteristics of the cleaned historical points exchange integrated data show the presence of data tailing characteristics, and the data partial autocorrelation characteristics show the presence of data truncation characteristics, a user points exchange prediction model is established based on the cleaned historical points exchange integrated data and the moving average model; When the data autocorrelation characteristics and partial autocorrelation characteristics of the historical point redemption integrated data after data cleaning both show the existence of data tailing characteristics, a user point redemption prediction model is established based on the historical point redemption integrated data after data cleaning and the autoregressive moving average model, wherein the data autocorrelation characteristics and partial autocorrelation characteristics of the historical point redemption integrated data after data cleaning do not simultaneously show data truncation characteristics.
4. The method according to claim 1, It is characterized in that The user points exchange prediction model is established based on the historical points exchange integration data after data cleaning and the autoregressive sliding average model, including: Based on the historical points exchange integration data after data cleaning and the autoregressive moving average model, determine the optimal order array of the autoregressive moving average model; Based on the autoregressive moving average model of the optimal order array, a user points redemption prediction model is established, wherein the autoregressive moving average model is: Y t,ARMA(p,q) =c+φ 1 y t-1 +φ 2 y t-2 +...+φ p y t-p +e t +θ 1 y t ′ -1 +θ 2 y t ′ -2 +...+θ q y t ′ -q , Y t,ARMA(p,q) represents the time series output term of the autoregressive moving average model with the optimal order set (p,q), where p and q are the autoregressive order and the moving average order respectively, c is the preset constant term, and φ 1 ,φ 2 ...φ p are the coefficients of the 1st to the pth autoregressive terms, y t-1 ,y t-2 ...y t-p are the autoregressive terms from t-1th to tpth, θ 1 ,θ 2 ...θ q are the coefficients of the 1st to qth moving average terms, y t ' -1 ,y t ' -2 ...y t ' -p are the moving average terms from t-1th to tqth, ε t is the preset noise term, and t is the total number of preset points integration periods in the historical points redemption integration data after data cleaning.
5. The method according to claim 4, It is characterized in that The method of determining the optimal order array of the autoregressive moving average model based on the historical points exchange integrated data after data cleaning and the autoregressive moving average model includes: Obtain multiple sets of candidate order arrays, and for any candidate order array, calculate the Akaike information value corresponding to the candidate order array based on the historical points exchange integration data after data cleaning, the autoregressive moving average model, the candidate order array and the Akaike information calculation formula; For each group of candidate order arrays corresponding to the Akaike information value, the candidate order array corresponding to the minimum Akaike information value is determined as the optimal order array of the autoregressive moving average model, wherein the Akaike information calculation formula is: AIC(i,j)=2k-2ln(L),k=i+j+1, AIC(i,j) is the Akaike information value corresponding to the alternative order array (i,j), i and j are both natural numbers, k is the parameter in the autoregressive moving average model, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical integral exchange integrated data after data cleaning and the alternative order array (i,j).
6. The method according to claim 4, It is characterized in that The method of determining the optimal order array of the autoregressive moving average model based on the historical points exchange integrated data after data cleaning and the autoregressive moving average model includes: Obtain multiple sets of candidate order arrays, and for any candidate order array, calculate the Bayesian information value corresponding to the candidate order array based on the historical points exchange integrated data after data cleaning, the autoregressive moving average model, the candidate order array and the Bayesian information calculation formula; For each group of candidate order arrays corresponding to the Bayesian information value, the candidate order array corresponding to the minimum Bayesian information value is determined as the optimal order array of the autoregressive moving average model, wherein the Bayesian information calculation formula is: BIC(m,l)=k ln(t)-2ln(L),k=m+l+1, BIC(m,l) is the Akaike information value corresponding to the alternative order array (m,l), m and l are both natural numbers, k is the parameter in the autoregressive moving average model, t is the total number of preset point integration periods in the historical point exchange integration data after data cleaning, and L is the maximum likelihood function value calculated by the autoregressive moving average model based on the historical point exchange integration data after data cleaning and the alternative order array (m,l).
7. The method according to any one of claims 1 to 6, It is characterized in that When the difference between the number of gifts to be exchanged corresponding to the expected exchange points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, sending gift distribution warning information to the preset recipient terminal includes: When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, the number of gifts to be allocated is obtained based on the difference between the number of gifts to be redeemed and the number of gifts retained in the current preset period, and gift distribution warning information is generated based on the type of gifts to be redeemed, the number of gifts to be allocated and the predicted future preset period, and the gift distribution warning information is sent to the preset recipient terminal.
8. A gift exchange warning device, It is characterized in that The device comprises: The points data acquisition module is used to acquire the user's historical points redemption data, wherein the historical points redemption data includes the redemption points and the redemption time; A points data integration module, used to sum up the redemption points redeemed within the same preset points integration period, and obtain historical points redemption integration data based on the summed total points redeemed in each of the multiple preset points integration periods, wherein the total points redeemed in each preset points integration period in the historical points redemption integration data are arranged in chronological order, and the time span of each preset points integration period is the same; A prediction model building module is used to clean the historical points exchange integrated data, and to build a user points exchange prediction model based on the historical points exchange integrated data after data cleaning and an autoregressive sliding average model; The gift redemption warning module is used to predict the user's expected redemption points in a future preset period based on the user points redemption prediction model. When the difference between the number of gifts to be redeemed corresponding to the expected redemption points and the number of gifts retained in the current preset period is greater than or equal to the preset risk value, a gift distribution warning message is sent to the preset recipient terminal.
9. A storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the method for gift redemption warning as described in any one of claims 1 to 7 is implemented.
10. A computer device comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, It is characterized in that When the processor executes the computer program, the method for gift redemption warning described in any one of claims 1 to 7 is implemented.
