User retention prediction processing method and apparatus, computer device, and storage medium

Abstract

The application relates to the technical field of Internet, and provides a user retention prediction processing method and device, computer equipment and a storage medium. The method comprises the following steps: obtaining cumulative user retention data corresponding to each historical time node at a respective subsequent time node according to user retention data possessed by each historical time node at the respective subsequent time node, and obtaining a change trend of the cumulative user retention data presented between adjacent time nodes based on the cumulative user retention data; and obtaining prediction user retention data corresponding to the historical time node at the prediction time node by using the change trend and the cumulative user retention data possessed by the corresponding historical time node at the previous time node of the prediction time node. The method can fully mine and utilize the retention data change law represented by the historical user retention data to accurately predict future user retention data, and improve the prediction accuracy and stability of the user retention data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet, in particular to a user retention prediction processing method and device, computer equipment and storage medium. BACKGROUND

[0002] With the development of Internet technology, analysis and processing technologies such as user access behavior and user retention prediction have emerged. In Internet technology, users who start using an application or an application service in a certain period of time and continue to use it after a period of time can be considered as retained users. The proportion of this part of users in the newly added users at that time is the retention rate. Generally, the retention rate can be counted according to time units such as days, weeks and months.

[0003] The prediction of user retention data provided in the current technology is usually to directly fit this year's user retention data with last year's user retention data. However, this method has the technical problem of low prediction accuracy of user retention data. SUMMARY

[0004] Therefore, it is necessary to provide a user retention prediction processing method and device, computer equipment and storage medium to solve the above technical problems.

[0005] A user retention prediction processing method, the method comprising:

[0006] obtaining user retention data possessed by each historical time node at its subsequent time node;

[0007] obtaining cumulative user retention data corresponding to each historical time node at its subsequent time node according to the user retention data possessed by each historical time node at its subsequent time node;

[0008] obtaining a change trend of cumulative user retention data between adjacent time nodes based on the cumulative user retention data corresponding to each historical time node at its subsequent time node;

[0009] obtaining predicted user retention data corresponding to the historical time node at the prediction time node according to the change trend and the cumulative user retention data possessed by the historical time node at the previous time node of the prediction time node.

[0010] A user retention prediction processing device, the device comprising:

[0011] a retention data obtaining module configured to obtain user retention data possessed by each historical time node at its subsequent time node;

[0012] an accumulation data obtaining module configured to obtain, according to the user retention data possessed by each historical time node at a respective subsequent time node, accumulation user retention data corresponding to each historical time node at a respective subsequent time node;

[0013] a change trend obtaining module configured to obtain a change trend of the accumulation user retention data between adjacent time nodes based on the accumulation user retention data corresponding to each historical time node at a respective subsequent time node;

[0014] a retention data prediction module configured to obtain predicted user retention data corresponding to the historical time node at a predicted time node according to the change trend and the accumulation user retention data possessed by the historical time node at a time node prior to the predicted time node.

[0015] A computer device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0016] obtaining user retention data possessed by each historical time node at a respective subsequent time node, obtaining, according to the user retention data possessed by each historical time node at a respective subsequent time node, accumulation user retention data corresponding to each historical time node at a respective subsequent time node, obtaining a change trend of the accumulation user retention data between adjacent time nodes based on the accumulation user retention data corresponding to each historical time node at a respective subsequent time node, and obtaining predicted user retention data corresponding to the historical time node at a predicted time node according to the change trend and the accumulation user retention data possessed by the historical time node at a time node prior to the predicted time node.

[0017] A computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the following steps:

[0018] obtaining user retention data possessed by each historical time node at a respective subsequent time node, obtaining, according to the user retention data possessed by each historical time node at a respective subsequent time node, accumulation user retention data corresponding to each historical time node at a respective subsequent time node, obtaining a change trend of the accumulation user retention data between adjacent time nodes based on the accumulation user retention data corresponding to each historical time node at a respective subsequent time node, and obtaining predicted user retention data corresponding to the historical time node at a predicted time node according to the change trend and the accumulation user retention data possessed by the historical time node at a time node prior to the predicted time node.

[0019] A computer program product or computer program comprises computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the steps of the above method.

[0020] The user retention prediction processing method, device, computer device and storage medium can obtain cumulative user retention data corresponding to each historical time node at each subsequent time node according to the user retention data of each historical time node at each subsequent time node, and obtain a change trend of the cumulative user retention data between adjacent time nodes based on the cumulative user retention data. Then, the predicted user retention data of the historical time node at the prediction time node is obtained from the change trend and the cumulative user retention data of the corresponding historical time node at the time node before the prediction time node. The scheme can obtain cumulative data corresponding to each subsequent time node by cumulating the user retention data of each historical time node, further obtain a change trend or change rule between adjacent time nodes from the cumulative data, and then predict the cumulative data of the prediction time node using the change trend or change rule to obtain the corresponding user retention data. The scheme can accurately predict future user retention data by fully mining and using the retention data change rule represented by multi-year user retention data, and improve the prediction accuracy and stability of the user retention data. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 An application environment diagram of the user retention prediction processing method in one embodiment;

[0022] Figure 2 A flowchart of the user retention prediction processing method in one embodiment;

[0023] Figure 3 A flowchart of the step of obtaining a change trend of cumulative user retention data between adjacent time nodes in one embodiment;

[0024] Figures 4(a) to 4(d) A schematic diagram of processing user retention data in some embodiments;

[0025] Figure 5 A flowchart of the step of obtaining the number of active users in one embodiment;

[0026] Figure 6 A flowchart of the user retention prediction processing method in another embodiment;

[0027] Figure 7A structural block diagram of a user retention prediction processing device in an embodiment;

[0028] Figure 8 An internal structural diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0029] For the purpose, technical solutions and advantages of the present application to be clearer, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0030] The user retention prediction processing method provided by the present application can be applied in an application environment as shown in Figure 1 The method can be executed by a server 120, which can be an independent server or a server cluster composed of multiple servers. The server 120 can provide background services for application programs, and different users can download and use application programs on their respective terminals. In the present application, the server 120 can predict user retention data corresponding to one or more prediction time nodes 13 in the future from the user retention data of each historical time node 11 at its respective subsequent time node. The user retention data can be a user retention rate or a specific number of user retentions. In the case of consistent data processing, in order to describe the present application clearly, the user retention rate will be used as the user retention data to describe the scheme in the subsequent part of the present application.

[0031] In addition, after predicting the user retention data corresponding to the prediction time node 13, the server 120 can further predict the daily active user number (DAU, Daily Active User) and total duration in the future, for example, in a certain year, based on the predicted user retention data. The predicted daily active user number and total duration in a certain year can provide reliable data support for related businesses in annual resource planning.

[0032] The user retention prediction processing method provided by the present application is described in detail below in combination with embodiments and corresponding drawings.

[0033] In an embodiment, as shown in Figure 2 A user retention prediction processing method is provided. Taking the server 120 in Figure 1 as an example, the method can include the following steps:

[0034] Step S201, obtaining user retention data of each historical time node at its respective subsequent time node;

[0035] Specifically, the time nodes can be divided or set according to a time granularity required for prediction, and the time granularity can be hours, days, weeks, months, years, etc. For example, from the release of the application, each day can be a time node.

[0036] The current time node and its previous time nodes can be historical time nodes. For example, taking months as the time granularity, if the current time node is December 2020, December 2020, November 2020, …, January 2020, …, etc. can be historical time nodes, and each historical time node has new users. As time goes on, the new users in each historical time node may be partially or completely retained in the subsequent time nodes. The server 120 can obtain the user retention rate of each historical time node in the subsequent time nodes according to the number of new users retained in the subsequent time nodes, that is, the server 120 can obtain the proportion of new users in each historical time node that is still retained in the subsequent time nodes. For the subsequent time nodes, for example, taking months as the time granularity, the subsequent time nodes of the historical time node of January 2020 can include 0, 1, 2, 3, …, and the user retention data of the historical time node in the subsequent time nodes means the retention rate of the new users generated in January 2020 in the subsequent 0, 1, 2, 3, … months.

[0037] In actual application, the daily active users or monthly active users can be composed of the users retained by each historical time node (for example, each historical month or each historical day). If the historical time node is calculated from the release date of the application, the daily active users of the application are mainly composed of the new users retained by each day in history. Based on the user retention rate of each historical time node in the subsequent time nodes, the application can predict the user retention rate of each historical time node in a certain or a certain period of prediction time nodes in the future. Based on the prediction result, the active user number of the prediction time node can be further calculated.

[0038] In step S202, the cumulative user retention data corresponding to each subsequent time node of each historical time node is obtained according to the user retention data of each historical time node in the respective subsequent time nodes.

[0039] This step is mainly to calculate the cumulative user retention data corresponding to each subsequent time node of each historical time node by the server 120, that is, for each historical time node, the cumulative user retention data corresponding to each subsequent time node is calculated. The cumulative user retention data corresponding to each subsequent time node is the sum of the user retention data corresponding to all the subsequent time nodes before the subsequent time node.

[0040] In an embodiment, step S202 can include:

[0041] The server 120 accumulates, for each historical time node, the user retention data that the historical time node has at each subsequent time node in time sequence in a forward direction, to obtain the cumulative user retention data corresponding to each historical time node at each subsequent time node.

[0042] For example, for a historical time node of September 2019, assuming that the subsequent time nodes include 0, 1 and 2, and the user retention data corresponding to the subsequent time nodes are 100%, 57.26% and 51.73% respectively, and the time sequence in a forward direction refers to the direction from the subsequent time node 0 to the subsequent time node 2, then the cumulative user retention data of the historical time node at the subsequent time node 0 is 100%, the cumulative user retention data of the historical time node at the subsequent time node 1 is 157.26%, and the cumulative user retention data of the historical time node at the subsequent time node 2 is 208.99%. In this embodiment, the server 120 can perform the same processing for each historical time node to obtain the cumulative user retention data corresponding to each historical time node at each subsequent time node.

[0043] Step S203: based on the cumulative user retention data corresponding to each historical time node at each subsequent time node, obtaining a change trend of the cumulative user retention data between adjacent time nodes;

[0044] This step is mainly to obtain, by the server 120, based on the cumulative user retention data corresponding to each historical time node at each subsequent time node, a change trend of the cumulative user retention data between each adjacent subsequent time node, which is fed back through the cumulative user retention data, and the change trend can reflect the change of the cumulative user retention data year by year, month by month or day by day in the overall historical period.

[0045] Step S204: obtaining, according to the change trend and the cumulative user retention data of the historical time node at the previous time node of the prediction time node, the prediction user retention data corresponding to the historical time node at the prediction time node.

[0046] In this step, the server 120 can calculate, according to the change trend of the cumulative user retention data between adjacent time nodes and in combination with the cumulative user retention data of each historical time node at the previous time node of the prediction time node, the prediction cumulative user retention data corresponding to each historical time node at the prediction time node, and the prediction user retention data corresponding to the prediction time node can be obtained by subtracting the prediction cumulative user retention data from the cumulative user retention data of the corresponding historical time node at the previous time node.

[0047] For example, according to the change trend of the accumulated user retention data between the subsequent 0 months and the subsequent 1 month, in combination with the accumulated user retention data of the historical time node December 2020 at the previous time node of the prediction time node January 2021, that is, December 2020, the server 120 can deduce that the predicted accumulated user retention rate corresponding to the prediction time node January 2021 of the historical time node December 2020 is, for example, 158.9%, and the server 120 can obtain the predicted user retention rate corresponding to the prediction time node January 2021 of December 2020 by subtracting the accumulated user retention data 100% corresponding to the historical time node December 2020 from the predicted accumulated user retention rate 158.9%. In this way, the server 120 can predict the predicted user retention rate corresponding to the prediction time node of each historical time node, and the server 120 can also estimate the user retention rate corresponding to the time node after the prediction time node according to the predicted user retention rate, that is, the server 120 can predict the user retention rate from January 2021 to December 2021 on the basis of predicting January 2021. In a specific application, the server 120 can estimate the user retention rate of the subsequent 365 days for each day in history. The larger the data volume is, the higher the estimation stability and accuracy are, and thus the more accurate and stable the estimated daily active user number is.

[0048] The above-mentioned user retention prediction processing method can obtain the accumulated user retention data corresponding to each historical time node at each subsequent time node according to the user retention data of each historical time node at each subsequent time node, and obtain the change trend of the accumulated user retention data between adjacent time nodes based on the accumulated user retention data. Then, the server 120 can obtain the predicted user retention data of the historical time node at the prediction time node according to the change trend and the accumulated user retention data of the corresponding historical time node at the previous time node of the prediction time node. In this scheme, the server 120 can obtain the accumulated data corresponding to each subsequent time node by accumulating the user retention data of each historical time node, and the server 120 can further obtain the change trend or change rule between adjacent time nodes by using the accumulated data. Then, the corresponding user retention data can be deduced by predicting the accumulated data of the prediction time node using the change trend or change rule. The change rule of the retention data represented by the multi-year user retention data can be fully mined and used to accurately predict the future user retention data, thereby improving the prediction accuracy and stability of the user retention data.

[0049] In one embodiment, the server 120 can obtain the change trend of the accumulated user retention data between adjacent time nodes in the following manner, as shown in FIG. 13 and in combination with FIG. 14. Figure 3 ​Figures 4(a) to 4(c) As explained above, the step S203 can include:

[0050] At step S301, for each subsequent time node, determine a data statistical range of the accumulated user retention data corresponding to the respective subsequent time node.

[0051] In the example, the historical time nodes T1-T4 can correspond to January 2020-April 2020, and the subsequent time nodes t0-t3 can correspond to the 0th, 1st, 2nd and 3rd months after the current time. In FIG. 4(a), X10-X40 represent the user retention rates of the historical time nodes at the respective subsequent time nodes, for example, the user retention rate X10 represents the user retention rate of the historical time node T1 at the subsequent time node t0. The server 120 can calculate the accumulated user retention rates of the historical time nodes T1-T4 at the respective subsequent time nodes t0-t3 by accumulating the historical time nodes T1-T4 respectively, as shown in FIG. 4(b), L10-L40 represent the accumulated user retention rates of the historical time nodes T1-T4 at the respective subsequent time nodes t0-t3.

[0052] In this step, the server 120 needs to determine, for each subsequent time node t1-t3, a data statistical range of the accumulated user retention data corresponding to the respective subsequent time node t1-t3. Specifically, the data statistical range determined by the server 120 for the subsequent time nodes t1-t3 includes a first data statistical range for each subsequent time node t1-t3 and a second data statistical range for the previous subsequent time node of each subsequent time node t1-t3, and the first data statistical range and the second data statistical range have the same size. As shown in FIG. 4(b), for the subsequent time node t1, the first data statistical range specifically refers to the column range occupied by L11, L21 and L31, and the second data statistical range specifically refers to the column range occupied by L10, L20 and L30; for the subsequent time node t2, the first data statistical range specifically refers to the column range occupied by L12 and L22, and the second data statistical range specifically refers to the column range occupied by L11 and L21; for the subsequent time node t3, the first data statistical range specifically refers to the column range occupied by L13, and the second data statistical range specifically refers to the column range occupied by L12.

[0053] At step S302, based on the accumulated user retention data corresponding to the first data statistical range and the second data statistical range of each historical time node, obtain first accumulated user retention data statistical information and second accumulated user retention data statistical information;

[0054] Specifically, as shown in FIG. 4(b), for the subsequent time node t1, the server 120 can count the accumulated user retention data corresponding to the column ranges occupied by L11, L21 and L31, and the column ranges occupied by L10, L20 and L30, to obtain the first accumulated user retention data statistical information and the second accumulated user retention data statistical information for the subsequent time node t1. Similarly, the server 120 can obtain the first accumulated user retention data statistical information and the second accumulated user retention data statistical information for the subsequent time nodes t2 and t3.

[0055] In some embodiments, for the subsequent time node t1, the server 120 can sum L11, L21 and L31 to obtain the first accumulated user retention data, and sum L10, L20 and L30 to obtain the second accumulated user retention data. Similarly, the first and second accumulated user retention data for the subsequent time nodes t2 and t3 can be calculated. That is, for each subsequent time node, the first accumulated user retention data statistical information can include the sum of the accumulated user retention data corresponding to the respective first data statistical range of each historical time node, and the second accumulated user retention data statistical information can include the sum of the accumulated user retention data corresponding to the respective second data statistical range of each historical time node.

[0056] Step S303: obtaining a change trend according to the first accumulated user retention data statistical information and the second accumulated user retention data statistical information.

[0057] In this step, for each subsequent time node, the server 120 can compare the obtained first and second accumulated user retention data statistical information to obtain the change trend of the accumulated user retention data for each subsequent time node.

[0058] Further, the above step S303 can specifically include: the server 120 obtains the change trend according to the ratio between the sum of the accumulated user retention data corresponding to the first data statistical range of each historical time node and the sum of the accumulated user retention data corresponding to the second data statistical range of each historical time node.

[0059] That is, for each subsequent time node, the first cumulative user retention data statistics information obtained by the server 120 includes the sum of the cumulative user retention data of each historical time node in the corresponding first data statistics range (referred to as first cumulative sum), and the second cumulative user retention data statistics information obtained by the server 120 includes the sum of the cumulative user retention data of each historical time node in the corresponding second data statistics range (referred to as second cumulative sum), so that the server 120 can calculate the ratio of the first cumulative sum and the second cumulative sum (referred to as cumulative sum ratio), as shown in FIG. 4(b), the server 120 can obtain the cumulative sum ratios of the subsequent time nodes t1 to t3 as F01, F12 and F23 respectively. The cumulative sum ratio can reflect the change trend information such as the change trend of the cumulative user retention data between the subsequent time nodes t0 and t1, the change trend of the cumulative user retention data between the subsequent time nodes t1 and t2, and the change trend of the cumulative user retention data between the subsequent time nodes t2 and t3, etc. Based on this, the server 120 can accurately and reliably calculate the cumulative user retention data of each historical time node at the prediction time node.

[0060] Further, in some embodiments, the server 120 can calculate the prediction user retention data of each historical time node at the prediction time node by the following way, and the above step S204 specifically includes:

[0061] The server 120 obtains the prediction cumulative user retention data of the historical time node at the prediction time node according to the product of the above ratio representing the change trend and the cumulative user retention data of the historical time node at the previous time node of the prediction time node; and the server 120 subtracts the prediction cumulative user retention data from the cumulative user retention data of the historical time node at the previous time node to obtain the prediction user retention data.

[0062] Referring to FIG. 4(b), without loss of generality, the server 120 can predict the accumulated user retention rate corresponding to the subsequent time node t1 for the historical time node T4, where the subsequent time node t1 is the prediction time node for the historical time node T4, and the server 120 can also predict the accumulated user retention rate corresponding to the subsequent time node t2 for the historical time node T3, where the subsequent time node t2 is the prediction time node for the historical time node T3. Specifically, referring to FIG. 4(b) and FIG. 4(c), after the server 120 obtains the change trend of the accumulated user retention data between the subsequent time nodes t0 and t1 represented by the accumulated ratio F01, the server 120 can multiply the accumulated ratio F01 by the accumulated user retention data L40 of the historical time node T4 at the previous time node t0 of the prediction time node t1, and the product is the predicted accumulated user retention data L41 of the historical time node T4 at the prediction time node t1.

[0063] By analogy, the server 120 can calculate the elements in all the blank cells in FIG. 4(b), that is, the predicted accumulated user retention data (L23 to L43) of each historical time node at the prediction time node. On this basis, the server 120 can restore the predicted accumulated user retention data as shown in FIG. 4(c) and the originally known accumulated user retention data to the corresponding predicted user retention data (X23 to X43) and the originally known user retention data (X10 to X40) in the reverse subtraction manner as shown in FIG. 4(d). For example, for the calculation of the predicted user retention data X41, the server 120 can subtract the accumulated user retention data L40 of the historical time node T4 at the previous time node t0 from the predicted accumulated user retention data L41 to obtain the predicted user retention data X41. By using the scheme provided in this embodiment, the server 120 can accurately and reliably calculate the predicted user retention data of each historical time node at one or more prediction time nodes.

[0064] After obtaining the predicted user retention rate of each historical time node at the prediction time node, the server 120 can calculate the number of active users at the prediction time node using the predicted user retention rate. In a specific business scenario, the daily active user number of the future, for example, is mainly composed of all retained new users and old users in history. For the first type of user, i.e., the new user, the server 120 can estimate the retention rate of all retained new users in history at each day, week, month, or year in the future in the manner described in the above embodiments, i.e., the server 120 can estimate the predicted user retention rate for the first type of user through the above manner. Further, the server 120 can also obtain the number of active users of the second type of user, i.e., the old user, at the prediction time node, based on which the server 120 can evaluate the number of active users at a certain prediction time node or a certain period of prediction time nodes. Specifically, in some embodiments, as shown in FIG. 6, the above method can further include the following steps: Figure 5

[0065] Step S501, the server 120 determines the number of the first type of users corresponding to each historical time node.

[0066] In this step, the server 120 can obtain the number of new users corresponding to each historical time node, for example, the number of new users joined in each month in history.

[0067] Step S502, obtaining the first active user number of the first type of user at the prediction time node according to the number of the first type of user at each historical time node and the corresponding predicted user retention rate.

[0068] Taking each day in history as the historical time node as an example, in this step, the server 120 can multiply the number of new users corresponding to each day in history with the corresponding predicted user retention rate at the prediction time node, so as to obtain the number of active users of the new user at the prediction time node, as the first active user number. In a specific application, the server 120 can calculate the number of all new users retained to December 31, 2021 according to the number of new users at each historical time node and the corresponding predicted user retention rate.

[0069] Step S503, the server 120 obtains the second active user number of the second type of user at the prediction time node.

[0070] In addition to the number of active users corresponding to the new user, in order to more reasonably and accurately predict the number of active users at the prediction time node, the server 120 also obtains the second active user number of the second type of user and the old user at the prediction time node. For example, in the case of an information application, the old user can be a user who has registered or used the information application before each historical time node.

[0071] ​In some embodiments, step S503 can include: determining, by the server 120, a second-class user base number corresponding to a selected time node before the prediction time node; obtaining, by the server 120, a user number decay historical trend of the second-class user, and obtaining a second active user number based on the user number decay historical trend and the second-class user base number.

[0072] In this embodiment, the server 120 can calculate the user number of the second-class user at the prediction time node from the selected time node according to the user number decay historical trend of the second-class user, as the second active user number. The user number decay historical trend refers to the decay trend of the number of the second-class user, i.e., the old user, before the selected time node. For example, the server 120 can predict the number of old users retained to December 31, 2021 in 2020. At this time, each day in 2021 can be regarded as a prediction time node, and the server 120 can select the daily active user number on December 31, 2020 (i.e., a selected time node) as the aforementioned second-class user base number. In the case of obtaining the user number decay historical trend, the server 120 can decay the second-class user base number according to the user number decay historical trend, i.e., the number of old users retained to each day in 2021 can be calculated, and thus the number of old users retained to December 31, 2021 can be obtained. For the calculation of the user number decay historical trend, the server 120 can take the daily active user number on December 31, 2020 as the basis, query the proportion of old users contained in the daily active user number one year ago, such as 52%, and obtain the proportion of new users in the daily active user number on December 31, 2020, such as 98%, so that the server 120 can calculate the proportion of old users retained from one year ago to December 31, 2020 as 46%, which is evenly distributed to each day in 2020, i.e., the attrition rate of old users in 2020, i.e., the user number decay historical trend, is about 99.87%, and the server 120 can calculate the number of old users in each day in 2021 from the daily active user number on December 31, 2020 as the aforementioned second active user number. In this way, the server 120 can apply the decay rule of old users in the previous year to the next year, for example, to accurately predict the active user number of old users corresponding to each day.

[0073] Step S504: obtaining the active user number corresponding to the prediction time node according to the first active user number and the second active user number.

[0074] In this step, the server 120 can add the first active user number and the second active user number to obtain the active user number corresponding to the prediction time node. For example, the server 120 can sum the first and second active user numbers of each day in 2021 to obtain the active user number of each day.

[0075] The scheme provided by the embodiments of the present application can be used for the server 120 to obtain the daily active user number DAU of each day in the future year based on the prediction of the retention quantity of new and old users in each day of the future year.

[0076] In one embodiment, a user retention prediction processing method is provided, which is executed by the server 120, as shown in the following. Figure 6 The method can include the following steps:

[0077] In step S601, the server 120 obtains the user retention rate of each historical time node in the respective subsequent time node.

[0078] In step S602, the server 120 obtains the cumulative user retention rate corresponding to each historical time node in the respective subsequent time node according to the user retention rate of each historical time node in the respective subsequent time node.

[0079] In step S603, the server 120 accumulates the user retention rate of each historical time node in the respective subsequent time node in time sequence in a forward direction to obtain the cumulative user retention rate corresponding to each historical time node in the respective subsequent time node.

[0080] In step S604, the server 120 determines a data statistical range of the cumulative user retention rate suitable for each subsequent time node.

[0081] The data statistical range includes a first data statistical range for each subsequent time node and a second data statistical range for the previous subsequent time node of each subsequent time node, and the first data statistical range and the second data statistical range have the same size.

[0082] In step S605, the server 120 obtains the change trend of the cumulative user retention rate between adjacent time nodes according to the ratio between the sum of the cumulative user retention rates of each historical time node in the first data statistical range and the sum of the cumulative user retention rates of each historical time node in the second data statistical range.

[0083] In step S606, the server 120 obtains the prediction cumulative user retention rate of each historical time node in the prediction time node according to the product of the above ratio and the cumulative user retention rate of each historical time node in the previous time node of the prediction time node.

[0084] Step S607: The server 120 subtracts each predicted cumulative user retention rate from the cumulative user retention rate of the corresponding historical time node at the previous time node to obtain the predicted user retention rate corresponding to each historical time node at the predicted time node.

[0085] In step S608, the server 120 obtains the first number of active users corresponding to the new users at the predicted time node based on the number of new users at each historical time node and the corresponding predicted user retention rate. Furthermore, the server 120 obtains the second number of active users corresponding to the old users at the predicted time node based on the historical trend of user number decay of the old users and the base number of old users.

[0086] In step S609 , the server 120 sums the first number of active users and the second number of active users to obtain the number of active users corresponding to the predicted time node.

[0087] In this embodiment of the present application, server 120 can perform a rolling estimate of new user retention rates over multiple years and continuously incorporate historical retention rate data during the estimation process. For each day, a subsequent retention rate estimate, for example, for 365 days, can be performed, with high estimation stability and accuracy. For the estimation of retention of existing users, server 120 can predict the retention rate for the next year based on the retention patterns of existing users in the previous year. Combining the two, the server 120 can ultimately estimate the number of daily active users (DAU) for each day of the next year. This facilitates internet business or application service providers to predict the number of daily active users (DAU) for each day of the next year based on their actual circumstances and formulate corresponding business service resource supply strategies.

[0088] It should be understood that, although the steps in the above flowchart are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the above flowchart may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0089] In one embodiment, Figure 7 As shown, a user retention prediction processing device is provided. The device can be a software module or a hardware module, or a combination of the two to form a part of a computer device. The device 700 specifically includes:

[0090] The retention data acquisition module 701 is used to obtain the user retention data of each historical time node at each subsequent time node;

[0091] The cumulative data obtaining module 702 is configured to obtain cumulative user retention data corresponding to each historical time node according to user retention data possessed by the historical time node at a respective subsequent time node;

[0092] The change trend obtaining module 703 is configured to obtain a change trend of cumulative user retention data between adjacent time nodes based on the cumulative user retention data corresponding to each historical time node at a respective subsequent time node.

[0093] The retention data prediction module 704 is configured to obtain predicted user retention data corresponding to the historical time node at a prediction time node according to the change trend and cumulative user retention data possessed by the historical time node at a time node preceding the prediction time node.

[0094] In one embodiment, the cumulative data obtaining module 702 is configured to, for each historical time node, accumulate the user retention data possessed by the respective subsequent time node in a forward time sequence to obtain cumulative user retention data corresponding to each historical time node at a respective subsequent time node.

[0095] In one embodiment, the change trend obtaining module 703 is configured to, for each subsequent time node, determine a data statistical range of cumulative user retention data suitable for each subsequent time node; wherein the data statistical range includes a first data statistical range for each subsequent time node and a second data statistical range for a preceding subsequent time node of each subsequent time node; the first data statistical range and the second data statistical range have the same size; based on cumulative user retention data corresponding to the first data statistical range and the second data statistical range of each historical time node, respectively, first cumulative user retention data statistical information and second cumulative user retention data statistical information are obtained; and the change trend is obtained according to the first cumulative user retention data statistical information and the second cumulative user retention data statistical information.

[0096] In one embodiment, the first cumulative user retention data statistical information includes a sum of cumulative user retention data corresponding to the first data statistical range of each historical time node; the second cumulative user retention data statistical information includes a sum of cumulative user retention data corresponding to the second data statistical range of each historical time node; and the change trend obtaining module 703 is configured to obtain the change trend according to a ratio between the sum of cumulative user retention data corresponding to the first data statistical range of each historical time node and the sum of cumulative user retention data corresponding to the second data statistical range of each historical time node.

[0097] In an embodiment, the retention data prediction module 704 is configured to obtain, according to a product of the ratio and the accumulated user retention data of the historical time node at a time node before the prediction time node, a prediction accumulated user retention data of the historical time node at the prediction time node; and obtain the prediction user retention data by subtracting the prediction accumulated user retention data from the accumulated user retention data of the historical time node at the time node before the prediction time node.

[0098] In an embodiment, the prediction user retention data comprises a prediction user retention rate of the first type of users; and the apparatus 700 can further comprise an active user number prediction module configured to determine a first type of user number corresponding to each of the historical time nodes; obtain a first active user number of the first type of users corresponding to the prediction time node according to the first type of user number of each of the historical time nodes and the prediction user retention rate corresponding to each of the historical time nodes; obtain a second active user number of the second type of users corresponding to the prediction time node; and obtain the active user number corresponding to the prediction time node according to the first active user number and the second active user number.

[0099] In an embodiment, the active user number prediction module is configured to determine a second type of user base number corresponding to a selected time node before the prediction time node; obtain a user number decay historical trend of the second type of users; the user number decay historical trend is about a decay trend of the number of the second type of users before the selected time node; and obtain the second active user number based on the user number decay historical trend and the second type of user base number.

[0100] The specific limitations of the user retention prediction processing apparatus can refer to the limitations of the user retention prediction processing method described above, and will not be repeated here. Each module in the user retention prediction processing apparatus can be realized by software, hardware and their combinations in whole or in part. Each module described above can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each module.

[0101] In an embodiment, a computer device is provided, which can be a server, and the internal structure diagram of the computer device can be as shown in FIG. 8. Figure 8As shown in the figure. The computer device 800 includes a processor 820, a memory and a network interface 840 connected through a system bus 810. Among them, the processor 820 of the computer device 800 is used to provide computing and control capabilities. The memory of the computer device 800 includes a non-volatile storage medium 8310, an internal memory 8320. The non-volatile storage medium 8310 stores an operating system 8311, a computer program 8312 and a database 8313. The internal memory 8320 provides an environment for the running of the operating system 8311 and the computer program 8312 in the non-volatile storage medium 8310. The database 8313 of the computer device 800 can be used to store user retention data, cumulative user retention data, change trends, predicted user retention data and active user data. The network interface 840 of the computer device 800 is used to communicate with external terminals through network connection. The computer program 8312 is executed by the processor 820 to implement a user retention prediction processing method.

[0102] Those skilled in the art can understand that, Figure 8 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0103] In one embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in the above method embodiments.

[0104] In one embodiment, a computer readable storage medium is provided, storing a computer program, which is executed by a processor to implement the steps in the above method embodiments.

[0105] In one embodiment, a computer program product or computer program is provided, which includes computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps in the above method embodiments.

[0106] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0107] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, but as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0108] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be pointed out that for those skilled in the art, without departing from the concept of the present application, some modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A user retention prediction processing method, characterized by, The method comprises: obtaining user retention data possessed by each historical time node at a respective subsequent time node; obtaining cumulative user retention data corresponding to each historical time node at a respective subsequent time node according to the user retention data possessed by each historical time node at the respective subsequent time node; for each subsequent time node, determining a data statistical range about cumulative user retention data suitable for each subsequent time node; wherein the data statistical range comprises a first data statistical range for each subsequent time node and a second data statistical range for a previous subsequent time node of each subsequent time node; the first data statistical range and the second data statistical range are of the same size; obtaining first cumulative user retention data statistical information and second cumulative user retention data statistical information based on the cumulative user retention data corresponding to the first data statistical range and the second data statistical range of each historical time node respectively; obtaining a change trend according to the first cumulative user retention data statistical information and the second cumulative user retention data statistical information; obtaining predicted user retention data corresponding to the historical time node at a predicted time node according to the change trend and the cumulative user retention data possessed by the historical time node at a previous time node of the predicted time node.

2. The method of claim 1, wherein, The method comprises: for each historical time node, accumulating the user retention data possessed at the respective subsequent time node in a forward time sequence to obtain cumulative user retention data corresponding to each historical time node at a respective subsequent time node.

3. The method of claim 1, wherein, The first cumulative user retention data statistical information comprises a sum of the cumulative user retention data corresponding to the first data statistical range of each historical time node; and the second cumulative user retention data statistical information comprises a sum of the cumulative user retention data corresponding to the second data statistical range of each historical time node. The method comprises: obtaining the change trend according to a ratio between the sum of the cumulative user retention data corresponding to the first data statistical range of each historical time node and the sum of the cumulative user retention data corresponding to the second data statistical range of each historical time node.

4. The method of claim 3, wherein, The method comprises: obtaining predicted cumulative user retention data corresponding to the historical time node at the predicted time node according to a product of the ratio representing the change trend and the cumulative user retention data possessed by the historical time node at the previous time node of the predicted time node. The predicted accumulated user retention data is subtracted from the accumulated user retention data that the historical time node has at the previous time node, to obtain the predicted user retention data.

5. The method according to any one of claims 1 to 4, characterized in that, The predicted user retention data comprises a predicted user retention rate for a first type of user; after the historical time node obtains the predicted user retention data corresponding to the prediction time node, the method further comprises: determining the number of first type of users corresponding to each historical time node; obtaining the first active user number of the first type of users corresponding to the prediction time node according to the number of first type of users that each historical time node has and the corresponding predicted user retention rate; obtaining the second active user number of the second type of users corresponding to the prediction time node; obtaining the active user number corresponding to the prediction time node according to the first active user number and the second active user number.

6. The method of claim 5, wherein, The method for obtaining the second active user number of the second type of users corresponding to the prediction time node comprises: determining the second type of user base number corresponding to a selected time node before the prediction time node of the second type of users; obtaining a user number decay historical trend for the second type of users; the user number decay historical trend is about the decay trend of the number of the second type of users before the selected time node; obtaining the second active user number based on the user number decay historical trend and the second type of user base number. 7.A user retention prediction processing apparatus, characterized by comprising: The device comprises: a retention data acquisition module for acquiring user retention data that each historical time node has at each subsequent time node; an accumulated data acquisition module for obtaining accumulated user retention data corresponding to each historical time node at each subsequent time node according to the user retention data that each historical time node has at each subsequent time node; a range determination module for determining a data statistical range of accumulated user retention data suitable for each subsequent time node for each subsequent time node; wherein the data statistical range comprises a first data statistical range for each subsequent time node and a second data statistical range for each subsequent time node. The first data statistical range and the second data statistical range are of the same size; a statistical information acquisition module for obtaining first accumulated user retention data statistical information and second accumulated user retention data statistical information based on the accumulated user retention data corresponding to the first data statistical range and the second data statistical range of each historical time node respectively; a trend acquisition module for obtaining a change trend according to the first accumulated user retention data statistical information and the second accumulated user retention data statistical information; a retention data prediction module for obtaining predicted user retention data corresponding to the prediction time node of the historical time node according to the change trend and the accumulated user retention data that the historical time node has at the previous time node of the prediction time node.

8. The apparatus of claim 7, wherein, The accumulation data obtaining module is further configured to, for each historical time node, accumulate the user retention data possessed by the respective subsequent time node in sequence and forwards to obtain the accumulated user retention data corresponding to each historical time node at the respective subsequent time node.

9. The apparatus of claim 7, wherein, The first accumulated user retention data statistical information comprises the sum of the accumulated user retention data corresponding to each historical time node in the first data statistical range; the second accumulated user retention data statistical information comprises the sum of the accumulated user retention data corresponding to each historical time node in the second data statistical range; and the trend obtaining module is further configured to obtain the change trend according to the ratio between the sum of the accumulated user retention data corresponding to each historical time node in the first data statistical range and the sum of the accumulated user retention data corresponding to each historical time node in the second data statistical range.

10. The apparatus of claim 9, wherein, The retention data prediction module is further configured to obtain the predicted accumulated user retention data corresponding to the historical time node at the prediction time node according to the product of the ratio representing the change trend and the accumulated user retention data possessed by the historical time node at a time node before the prediction time node; and obtain the prediction user retention data by subtracting the accumulated user retention data possessed by the historical time node at the time node before the prediction time node from the predicted accumulated user retention data.

11. The apparatus of any one of claims 7 to 10, wherein, The prediction user retention data comprises a prediction user retention rate for a first type of user; the device further comprises an active user number prediction module configured to determine the number of the first type of user corresponding to each historical time node; obtain the first active user number corresponding to the prediction time node for the first type of user according to the number of the first type of user possessed by each historical time node and the corresponding prediction user retention rate; obtain the second active user number corresponding to the prediction time node for a second type of user; and obtain the active user number corresponding to the prediction time node according to the first active user number and the second active user number.

12. The apparatus of claim 11, wherein, The active user number prediction module is further configured to determine the second type of user base number corresponding to a selected time node before the prediction time node for the second type of user; obtain a user number decay historical trend for the second type of user; and the user number decay historical trend is a decay trend of the number of the second type of user possessed before the selected time node; obtain the second active user number based on the user number decay historical trend and the second type of user base number.

13. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to implement the method in any one of claims 1 to 6.

14. A computer readable storage medium storing a computer program, wherein the computer program comprises program instructions configured to cause a processor to perform the method according to any one of claims 1 to 13. The computer program is executed by the processor to implement the method in any one of claims 1 to 6.

15. A computer program product comprising computer instructions, characterized in that, The computer program is executed by the processor to implement the method in any one of claims 1 to 6.

Images