Subscription service life cycle value prediction method based on multi-dimensional data analysis

Abstract

The invention provides a subscription service life cycle value prediction method based on multi-dimensional data analysis, and the method comprises the steps: building a multi-dimensional churn rate prediction model through obtaining user behavior data, and the historical income data and behavior data of a subscription service user, and carrying out the calculation of a user churn probability parameter, and constructing a subscription service life cycle value prediction model based on the obtained user loss probability parameter, historical income data and behavior data, and dynamically updating the subscription service life cycle value prediction model based on the obtained user behavior data and the change of the loss probability parameter in combination with the influence of the user behavior change and the loss risk thereof. And obtaining a dynamic user life cycle value prediction result. By combining multi-dimensional analysis of user behavior data, a wastage rate prediction model and a life cycle value prediction model are established, and a prediction result is dynamically adjusted, so that prediction precision and real-time performance are remarkably improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of data processing, and in particular to a subscription service life cycle value prediction method based on multidimensional data analysis. Background Art

[0002] In the development of subscription services, user churn is a problem faced by operators. In the existing technology, subscription service providers usually conduct preliminary evaluation of user value through simple user behavior statistics, such as login frequency, usage time, etc. The preliminary evaluation of user value fails to fully consider the multi-dimensional changes in user behavior and the dynamic changes in its churn risk, and has the following deficiencies:

[0003] 1. Most existing lifetime value prediction models rely on static data analysis methods, which makes them unable to reflect the dynamic changes of user behavior and its impact on churn risk in real time.

[0004] 2. Although some models have introduced users’ historical income data and some behavioral data, these models have not combined users’ multi-dimensional behavioral changes and churn probability parameters, resulting in low accuracy of their prediction results. In particular, in scenarios with high diversity of user behaviors, the applicability of traditional models is greatly reduced. Summary of the invention

[0005] The purpose of the embodiment of the present invention is to provide a subscription service life cycle value prediction method based on multi-dimensional data analysis. By implementing the embodiment of the present invention, a multi-dimensional churn rate prediction model is constructed by acquiring user behavior data, and the prediction results of the user's life cycle value are dynamically adjusted in combination with the user's historical income data and behavior changes. Compared with the existing technology, this method significantly improves the accuracy and real-time performance of life cycle value prediction, and can help subscription service providers more accurately evaluate user value, optimize business decisions, and thus improve overall operational efficiency.

[0006] To achieve the purpose, an embodiment of the present invention provides a subscription service life cycle value prediction method based on multi-dimensional data analysis, including:

[0007] Obtain user behavior data, build a multi-dimensional churn rate prediction model, and calculate the user churn probability parameters;

[0008] Obtain historical revenue data and behavior data of subscription service users, and build a subscription service life cycle value prediction model based on the obtained user churn probability parameters, historical revenue data and behavior data;

[0009] Based on the changes in the obtained user behavior data and the churn probability parameters, combined with the impact of user behavior changes and their churn risk, the subscription service life cycle value prediction model is dynamically updated to obtain dynamic user life cycle value prediction results.

[0010] In some specific implementations, user behavior data is obtained, a multi-dimensional churn rate prediction model is constructed, and a user churn probability parameter is calculated, and the formula is:

[0011]

[0012] Where P churn，t represents the probability of user churn in the tth time period, e represents the exponential decay function in the churn rate model, X i represents the user's i-th behavior data, α i Represents the behavior data X of each user i The relevant weights, m represents the amount of user behavior data, and t represents the time period used for calculation.

[0013] Among them, α i Based on the analysis of historical data, historical data includes user behavior feature data and churn results. By collecting user behavior feature data and churn results and using the Pearson correlation coefficient to calculate the correlation between each behavior feature and the churn rate, and by calculating and analyzing the user's login frequency, usage time, function usage and user evaluation characteristics, their impact on user churn is determined.

[0014] In some specific implementations, the user churn probability parameter is associated with the result of a subscription service lifetime value prediction model, and the user churn probability parameter is used to adjust the expectation of the user lifetime value.

[0015] In some specific implementations, the user behavior data includes the login frequency, usage duration, function usage ratio and service evaluation during the user's subscription service period.

[0016] In some specific implementations, the multi-dimensional churn rate prediction model predicts the churn probability parameters of users based on their behavioral characteristic data during the subscription service period, and feeds the churn probability parameters back into the subscription service lifetime value prediction model as a parameter for adjusting the value of the subscription service.

[0017] In some specific implementations, historical income data and behavior data of subscription service users are obtained, and a subscription service life cycle value prediction model is constructed based on the obtained user churn probability parameter, historical income data and behavior data. The calculation formula is:

[0018]

[0019] Where LTV represents the lifetime value prediction result of subscription service users, R t represents the user’s income in time period t, P churn，trepresents the probability parameter of user churn in the tth time period, d represents the discount rate, n represents the length of the predicted cycle, and t represents the time period used for calculation.

[0020] It should be noted that the LTV mentioned above is spelled as Lifetime Value, which is a business indicator of subscription services.

[0021] In some specific implementations, based on the obtained user behavior data and the changes in the churn probability parameters, combined with the impact of the user behavior changes and their churn risk, the subscription service life cycle value prediction model is dynamically updated to obtain a dynamic user life cycle value prediction result, and the update formula is:

[0022]

[0023] Where LTV 动态 represents the user lifetime value prediction result after dynamic update, R t represents the user's income in the tth time period, f(t) represents the function of user behavior changing over time, β represents the coefficient of the impact of behavior change on the risk of churn, and P churn，t represents the probability of user churn in the tth time period, d represents the discount rate, and t represents the time period used for calculation.

[0024] Among them, LTV 动态 By acquiring user behavior data in real time and combining the churn probability and behavior fluctuation function to dynamically adjust the user's lifetime value prediction, this dynamic update mechanism can promptly reflect the impact of changes in user behavior on their future revenue, ensuring that the prediction model has the ability to adjust when user behavior fluctuates.

[0025] In some specific embodiments, it also includes:

[0026] Based on the historical revenue data of subscription service users, the historical lifetime value of users is calculated using the following formula:

[0027]

[0028] In the formula, R t represents the user’s income in time period t, w t represents the weight of time period t, n represents the number of time periods for historical data analysis, and LTV 历史 represents the historical lifetime value of a user, and t represents the time period used for calculation.

[0029] Among them, LTV 历史The calculation is based on the user's income data in the past several time periods. Higher weights are given to more recent time periods to reflect the impact of the user's recent behavior on their long-term value. The long-term lifetime value of the user is obtained by taking a weighted average of these data.

[0030] In some specific embodiments, it also includes:

[0031] Based on the historical life cycle value of the user and the dynamically updated user life cycle value prediction results, a fusion life cycle value prediction model is constructed, and its fusion formula is:

[0032] LRV 融合 =λ LTV 历史 +(1-λ) LTV 动态

[0033] In the formula, λ represents the balance coefficient, LTV 历史 Represents the historical lifetime value of a user, LTV 动态 Represents the dynamically updated user lifetime value prediction result, LTV 融合 Indicates combined LTV 历史 and LTV 动态 The comprehensive prediction results.

[0034] The specific value range of λ is (0, 1).

[0035] In some specific implementations, by introducing λ, the model can 历史 and LTV 动态 When user behavior fluctuates drastically, λ dynamically adjusts the weights of historical data and real-time data.

[0036] The fusion model not only considers the stability of long-term data, but also reflects the changes in user behavior in real time, thus solving the technical problem that conventional models are difficult to strike a balance between processing long-term data and real-time data. The fusion model improves the flexibility and accuracy of prediction by combining data from multiple behavioral feature dimensions with user churn risk and real-time behavioral changes, so that the present invention can adapt to a variety of subscription service scenarios, especially in an environment with complex user behavior and volatile market conditions, and demonstrates accurate prediction results.

[0037] In some specific embodiments, it also includes:

[0038] Generate a user preference feature model based on the user's multi-dimensional behavior data, specifically:

[0039]

[0040] In the formula, X irepresents the user's i-th behavior data, α i Represents the behavior data X of each user i The relevant weight, m, represents the total number of user behavior features.

[0041] Among them, by calculating the user preference score, this model can quantify the user's preferences in different dimensions, thereby providing a comprehensive evaluation index for subsequent personalized adjustments. The higher the preference score, the stronger the user's dependence on the platform, and vice versa, the greater the risk of user loss.

[0042] In some specific embodiments, it also includes:

[0043] According to the user preference feature model, the fusion life cycle value prediction result is personalized adjusted. After the user preference score is obtained, it is used as the adjustment factor of the fusion life cycle value prediction model. The adjustment factor is defined as follows:

[0044]

[0045] In the formula, δ represents the adjustment factor, and the user preference score max represents the maximum value of the user preference score, which is used to standardize the preference score;

[0046] Based on the fusion life cycle value prediction results, an adjustment factor is introduced to personalize the fusion life cycle value prediction results through the preference feature model. Specifically:

[0047] LTV 个性化 =LTV 融合 ×δ

[0048] Where LTV 个性化 Represents the predicted lifetime value after personalization adjustment, LTV 融合 Indicates combined LTV 历史 and LTV 动态 The comprehensive prediction result of , δ represents the adjustment factor.

[0049] Among them, through this adjustment, the personalized adjusted LTV prediction can more accurately reflect the long-term value of users on the platform. When the user preference score is high, the δ value will increase accordingly, thereby increasing the LTV prediction result; when the user preference score is low, δ will decrease, making the LTV prediction result more in line with the actual churn risk, making differentiated predictions for high-value users and potential churn users, helping subscription service platforms adopt corresponding personalized strategies, thereby improving overall business results.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. The present invention establishes a churn rate prediction model and a lifetime value prediction model by combining multi-dimensional analysis of user behavior data, and dynamically adjusts the prediction results, thereby significantly improving the prediction accuracy and real-time performance. Compared with conventional technologies, the present invention not only considers the user's historical income data, but also considers the real-time changes in user behavior, ensuring that the prediction results can promptly reflect the impact of user behavior fluctuations, thereby improving the accuracy of subscription service user value assessment.

[0052] 2. By introducing a fusion lifecycle value prediction model, the present invention combines the advantages of historical data and real-time data. The model dynamically adjusts the weights through a balance coefficient, so that it can not only retain the long-term stability of historical data, but also respond to changes in user behavior in real time, adapt to the complex and changeable environment of user behavior, and improve the robustness of the prediction model in the face of market fluctuations and user behavior diversity. It is suitable for a variety of subscription service scenarios.

[0053] 3. Through the prediction of churn rate and lifetime value based on multi-dimensional data, the present invention can provide subscription service providers with more accurate user value assessment tools. The real-time dynamically adjusted prediction results can help operators to promptly identify high-risk churn users, thereby adopting effective retention strategies, optimizing the operational decisions of subscription services, and ultimately improving overall business efficiency and revenue.

[0054] In addition, the above-described invention summary does not list all the features required for the embodiments of the present invention, and other combinations of these feature groups may also constitute embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the background technology, the drawings required for use in the embodiments of the present invention or the background technology will be described below.

[0056] Figure 1 This is a flowchart of a subscription service life cycle value prediction method based on multi-dimensional data analysis provided by this application. DETAILED DESCRIPTION

[0057] In order to make the technical means, creative features, objectives and effects achieved by the embodiments of the present invention easy to understand, the embodiments of the present invention are further described below in combination with diagrams and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the embodiments of the present invention and are not used to limit the embodiments of the present invention.

[0058] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0059] See also Figure 1 , Figure 1A flowchart of a subscription service lifecycle value prediction method based on multidimensional data analysis provided in an embodiment of the present application.

[0060] In this embodiment, the subscription service life cycle value prediction method based on multi-dimensional data analysis includes step S101, step S102 and step S103.

[0061] Step S101: Obtain user behavior data, build a multi-dimensional churn rate prediction model, and calculate user churn probability parameters.

[0062] Here, we obtain user behavior data, build a multi-dimensional churn rate prediction model, and calculate the user churn probability parameter. The formula is:

[0063]

[0064] Where P churn，t represents the probability of user churn in the tth time period, e represents the exponential decay function in the churn rate model, X i represents the user's i-th behavior data, α i Represents the behavior data X of each user i The relevant weights, m represents the amount of user behavior data, and t represents the time period used for calculation.

[0065] Among them, α i Based on the analysis of historical data, historical data includes user behavior feature data and churn results. By collecting user behavior feature data and churn results and using the Pearson correlation coefficient to calculate the correlation between each behavior feature and the churn rate, and by calculating and analyzing the user's login frequency, usage time, function usage and user evaluation characteristics, their impact on user churn is determined.

[0066] Among them, the user churn probability parameter is associated with the result of the subscription service lifetime value prediction model, and the user churn probability parameter is used to adjust the expectation of the user lifetime value.

[0067] Among them, user behavior data includes login frequency, usage time, function usage ratio and service evaluation during the user's subscription period.

[0068] Among them, the multi-dimensional churn rate prediction model predicts the churn probability parameters based on the user's behavioral characteristic data during the subscription service period, and feeds the churn probability parameters back to the subscription service life cycle value prediction model as a parameter for adjusting the subscription service value.

[0069] Among them, before calculating the churn rate, user behavior data is collected from the interaction between users and subscription services. The user behavior data comes from the interaction between users and the platform, covering the user's login frequency, usage time, function usage and service evaluation. User behavior data is collected in real time through background monitoring to form a data set for subsequent churn rate prediction.

[0070] After data collection, the raw data needs to be extracted into key features through the data analysis process. The feature extraction process converts user behavior into digital features that can be understood by the model. User behavior features are extracted through preprocessing and analysis methods. User behavior features are the key input in the data analysis process.

[0071] After extracting user behavior features, the user behavior features are input into the churn rate prediction model. The churn rate prediction model will calculate the contribution of each feature to the churn rate through feature weights, and finally derive the user's churn probability parameter.

[0072] After the churn rate is calculated, it will be used as an input to the subscription service lifetime value prediction model to dynamically adjust the user lifetime value prediction results. The higher the churn rate, the lower the possibility that the user will remain subscribed in the future time period, and the future revenue expectations will be reduced accordingly.

[0073] User behavior data will be continuously monitored and updated. Based on the latest user behavior data, the user churn rate will be recalculated regularly and the lifetime value prediction model will be adjusted.

[0074] In this way, a churn rate prediction model and a lifetime value prediction model are established, and the prediction results are dynamically adjusted, which significantly improves the prediction accuracy and real-time performance. Compared with conventional technologies, the present invention not only takes into account the user's historical income data, but also takes into account the real-time changes in user behavior, ensuring that the prediction results can promptly reflect the impact of user behavior fluctuations, thereby improving the accuracy of subscription service user value assessment.

[0075] Step S102: Obtain historical income data and behavior data of subscription service users, and build a subscription service life cycle value prediction model based on the obtained user churn probability parameters, historical income data and behavior data.

[0076] Here, the historical income data and behavior data of subscription service users are obtained, and a subscription service life cycle value prediction model is constructed based on the obtained user churn probability parameters, historical income data and behavior data. The calculation formula is:

[0077]

[0078] Where LTV represents the lifetime value prediction result of subscription service users, R t represents the user’s income in time period t, Pchurn，t represents the probability parameter of user churn in the tth time period, d represents the discount rate, n represents the length of the predicted cycle, and t represents the time period used for calculation.

[0079] In this way, the user's historical income data, behavioral data and churn probability are comprehensively considered to achieve an accurate prediction of the user's lifetime value. By integrating multi-dimensional data, the constructed prediction model not only reflects the user's past behavior and consumption trends, but also can respond to changes in user churn risks in a timely manner, thereby achieving a comprehensive and dynamic assessment of the user's future value.

[0080] Step S103: Based on the obtained user behavior data and the changes in the churn probability parameters, combined with the impact of the user behavior changes and their churn risks, the subscription service life cycle value prediction model is dynamically updated to obtain a dynamic user life cycle value prediction result.

[0081] Here, combined with the impact of user behavior changes and their churn risk, the subscription service life cycle value prediction model is dynamically updated to obtain the dynamically updated user life cycle value prediction result. The update formula is:

[0082]

[0083] Where LTV 动态 represents the user lifetime value prediction result after dynamic update, R t represents the user's income in the tth time period, f(t) represents the function of user behavior changing over time, β represents the coefficient of the impact of behavior change on the risk of churn, and P churn，t represents the probability of user churn in the tth time period, d represents the discount rate, and t represents the time period used for calculation.

[0084] In this way, LTV 动态 By acquiring user behavior data in real time and combining the churn probability and behavior fluctuation function to dynamically adjust the user's lifetime value prediction, this dynamic update mechanism can promptly reflect the impact of changes in user behavior on their future revenue, ensuring that the prediction model has the ability to adjust when user behavior fluctuates.

[0085] Here, step 103 also includes:

[0086] Based on the historical revenue data of subscription service users, the historical lifetime value of users is calculated using the following formula:

[0087]

[0088] In the formula, R t represents the user’s income in time period t, w trepresents the weight of time period t, n represents the number of time periods for historical data analysis, and LTV 历史 represents the historical lifetime value of a user, and t represents the time period used for calculation.

[0089] Among them, LTV 历史 The calculation is based on the user's income data in the past several time periods. Higher weights are given to more recent time periods to reflect the impact of the user's recent behavior on their long-term value. The long-term lifetime value of the user is obtained by taking a weighted average of these data.

[0090] Here, the model needs to consider the stability of long-term data and the real-time changes in user behavior.

[0091] To this end, step 103 also includes:

[0092] Based on the historical life cycle value of the user and the dynamically updated user life cycle value prediction results, a fusion life cycle value prediction model is constructed, and its fusion formula is:

[0093] LTV 融合 =λ LTV 历史 +(1-λ) LTV 动态

[0094] In the formula, λ represents the balance coefficient, LTV 历史 Represents the historical lifetime value of a user, LTV 动态 Represents the dynamically updated user lifetime value prediction result, LTV 融合 Indicates combined LTV 历史 and LTV 动态 The comprehensive prediction results.

[0095] The specific value range of λ is (0, 1).

[0096] In this way, the fusion model not only considers the stability of long-term data, but also reflects the changes in user behavior in real time, thereby solving the technical problem that conventional models are difficult to strike a balance between processing long-term data and real-time data. The fusion model improves the flexibility and accuracy of prediction by combining data from multiple behavioral feature dimensions with user churn risk and real-time behavioral changes, so that the present invention can adapt to a variety of subscription service scenarios, especially in an environment with complex user behavior and volatile market conditions, and shows good prediction results.

[0097] It should be noted that step S103 also includes:

[0098] Generate a user preference feature model based on the user's multi-dimensional behavior data, specifically:

[0099]

[0100] In the formula, X i represents the user's i-th behavior data, α i Represents the behavior data X of each user i The relevant weight, m, represents the total number of user behavior features.

[0101] It should be further explained that by calculating the user preference score, this model can quantify the user's preferences in different dimensions, thereby providing a comprehensive evaluation index for subsequent personalized adjustments.

[0102] It should be further explained that the fusion life cycle value prediction result is personalized according to the user preference feature model. After the user preference score is obtained, it is used as the adjustment factor of the fusion life cycle value prediction model. The adjustment factor is defined as follows:

[0103]

[0104] In the formula, δ represents the adjustment factor, and the user preference score max represents the maximum value of the user preference score, which is used to standardize the preference score;

[0105] Based on the fusion life cycle value prediction results, an adjustment factor is introduced to personalize the fusion life cycle value prediction results through the preference feature model. Specifically:

[0106] LTV 个性化 =LTV 融合 ×δ

[0107] Where LTV 个性化 Represents the predicted lifetime value after personalization adjustment, LTV 融合 Indicates combined LTV 历史 and LTV 动态 The comprehensive prediction result of , δ represents the adjustment factor.

[0108] It is further necessary to explain that through this adjustment, the personalized adjusted LTV prediction can more accurately reflect the long-term value of users on the platform. When the user preference score is high, the δ value will increase accordingly, thereby increasing the LTV prediction result; when the user preference score is low, δ will decrease, making the LTV prediction result more in line with the actual churn risk, making differentiated predictions for high-value users and potential churn users, helping subscription service platforms adopt corresponding personalized strategies, thereby improving overall business results.

[0109] It should be understood that "one embodiment" or "an embodiment" or "a feasible implementation" or "some implementations" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the embodiments of the present invention. Therefore, "in one embodiment" or "in an embodiment" or "a feasible implementation" or "some implementations" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. Those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required for the embodiments of the present invention.

[0110] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by instructions. These instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0111] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0112] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0113] The above is only a specific implementation of the present invention, but the protection scope of the embodiments of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the embodiments of the present invention, which should be included in the protection scope of the embodiments of the present invention. Therefore, the protection scope of the embodiments of the present invention shall be based on the protection scope of the claims.

Claims

1. A subscription service life cycle value prediction method based on multidimensional data analysis, characterized in that: include: Obtain user behavior data, build a multi-dimensional churn rate prediction model, and calculate the user churn probability parameters; Obtain historical revenue data and behavior data of subscription service users, and build a subscription service life cycle value prediction model based on the obtained user churn probability parameters, historical revenue data and behavior data; Based on the changes in the obtained user behavior data and the churn probability parameters, combined with the impact of user behavior changes and their churn risk, the subscription service life cycle value prediction model is dynamically updated to obtain dynamic user life cycle value prediction results.

2. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 1 is characterized in that: Obtain user behavior data, build a multi-dimensional churn rate prediction model, and calculate the user churn probability parameter. The formula is: Where P churn，t represents the probability of user churn in the tth time period, e represents the exponential decay function in the churn rate model, X i represents the user's i-th behavior data, α i Represents the behavior data X of each user i The relevant weights, m represents the amount of user behavior data, and t represents the time period used for calculation.

3. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 2 is characterized in that: The user churn probability parameter is associated with the result of the subscription service lifetime value prediction model, and the user churn probability parameter is used to adjust the expectation of the user lifetime value.

4. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 2 is characterized in that: User behavior data includes login frequency, usage duration, function usage ratio and service evaluation during the user's subscription period.

5. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 2 is characterized in that: Obtain the historical income data and behavior data of subscription service users, and build a subscription service life cycle value prediction model based on the obtained user churn probability parameters, historical income data and behavior data. The calculation formula is: Where LTV represents the lifetime value prediction result of subscription service users, R t represents the user’s income in time period t, P churn，t represents the probability parameter of user churn in the tth time period, d represents the discount rate, n represents the length of the predicted cycle, and t represents the time period used for calculation.

6. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 5 is characterized in that: Based on the changes in the obtained user behavior data and the churn probability parameters, combined with the impact of user behavior changes and their churn risk, the subscription service life cycle value prediction model is dynamically updated to obtain the dynamic user life cycle value prediction result. The update formula is: Where LTV 动态 represents the user lifetime value prediction result after dynamic update, R t represents the user's income in the tth time period, f(t) represents the function of user behavior changing over time, β represents the coefficient of the impact of behavior change on the risk of churn, and P churn，t represents the probability of user churn in the tth time period, d represents the discount rate, and t represents the time period used for calculation.

7. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 6, characterized in that: Also includes: Based on the historical revenue data of subscription service users, the historical lifetime value of users is calculated using the following formula: In the formula, R t represents the user’s income in time period t, w t represents the weight of time period t, n represents the number of time periods for historical data analysis, and LTV 历史 represents the historical lifetime value of a user, and t represents the time period used for calculation.

8. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 7 is characterized in that: Also includes: Based on the historical life cycle value of the user and the dynamically updated user life cycle value prediction results, a fusion life cycle value prediction model is constructed, and its fusion formula is: LTV 融合 =λ·LTV 历史 +(1-λ)·LTV 动态 In the formula, λ represents the balance coefficient, LTV 历史 Represents the historical lifetime value of a user, LTV 动态 Represents the dynamically updated user lifetime value prediction result, LTV 融合 Indicates combined LTV 历史 and LTV 动态 The comprehensive prediction results.

9. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 8, characterized in that: Also includes: Generate a user preference feature model based on the user's multi-dimensional behavior data, specifically: Where, X i represents the user's i-th behavior data, α i Represents the behavior data X of each user i The relevant weight, m, represents the total number of user behavior features.

10. The subscription service life cycle value prediction method based on multidimensional data analysis according to claim 9, characterized in that: Also includes: According to the user preference feature model, the fusion life cycle value prediction result is personalized adjusted. After the user preference score is obtained, it is used as the adjustment factor of the fusion life cycle value prediction model. The adjustment factor is defined as follows: In the formula, δ represents the adjustment factor, and the user preference score max Indicates the maximum value of the user preference score, which is used to standardize the preference score; Based on the fusion life cycle value prediction results, an adjustment factor is introduced to personalize the fusion life cycle value prediction results through the preference feature model. Specifically: ATV 个性化 =LTV 融合 ×δ Where LTV 个性化 Represents the predicted lifetime value after personalization adjustment, LTV 融合 Indicates combined LTV 历史 and LTV 动态 The comprehensive prediction result of , δ represents the adjustment factor.

Images