CN112686718B - Method and device for acquiring user loss reason, computer equipment and storage medium - Google Patents
Method and device for acquiring user loss reason, computer equipment and storage medium Download PDFInfo
- Publication number
- CN112686718B CN112686718B CN202110293420.3A CN202110293420A CN112686718B CN 112686718 B CN112686718 B CN 112686718B CN 202110293420 A CN202110293420 A CN 202110293420A CN 112686718 B CN112686718 B CN 112686718B
- Authority
- CN
- China
- Prior art keywords
- user
- loss
- value
- churn
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The application relates to a method and a device for acquiring a user loss reason, computer equipment and a storage medium. The method comprises the following steps: acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user; calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn; calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features; and acquiring the user loss reason according to the loss value ratio of each user characteristic. By adopting the method, the loss reason of the user can be accurately acquired.
Description
Technical Field
The present application relates to the field of computer application technologies, and in particular, to a method and an apparatus for obtaining a cause of user churn, a computer device, and a storage medium.
Background
With the development of science and technology, more and more application programs appear in order to meet the requirements of people on work and life. For developers, it is generally desirable that the developed application has high adhesion with users. After an application is online, a user is often lost. Therefore, the early warning analysis of user loss becomes an important ring in the user relationship management; the early warning analysis of the user loss firstly needs to acquire the reason of the user loss.
However, in the conventional technology, machine learning algorithms used in the user loss early warning field generally have a problem of poor interpretability, that is, the loss early warning model generally only can give a result of whether a user will be lost, and cannot accurately obtain the loss reason of each user, which increases difficulty for subsequent user saving. Therefore, how to accurately obtain the loss reason of the user becomes an urgent problem to be solved.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a method, an apparatus, a computer device, and a storage medium for obtaining a user churn cause, which can accurately obtain a churn cause of a user.
A method for acquiring a user churn reason, the method comprising:
acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn;
calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
and acquiring the user loss reason according to the loss value ratio of each user characteristic.
In one embodiment, the method further comprises the following steps: inputting target user data into a pre-trained lost user judgment model, and outputting lost user data;
the loss user judgment model is obtained by training according to sample user data and a loss label of a sample user; the sample user data comprises not less than one user characteristic of a sample user; the attrition label is a parameter used to characterize whether the sample user is an attrition user.
In one embodiment, the churn user data is substituted into a preset churn value calculation formula to obtain a churn value of each user characteristic;
the loss value calculation formula is as follows:
wherein "! "represents factorial," | "represents the number of elements contained in a set, F represents a set of all user characteristics,
indicating the removal of the ith bit from the F setThe feature set left after characterization, S represents a subset of F,
representing an attrition user judgment model trained after adding the ith feature to the feature subset S,
an attrition user judgment model trained based on the user feature subset S is represented.
In one embodiment, replacing the data less than 0 in the churn value of each user characteristic with 0, and calculating the sum of the churn values of each user characteristic as a total churn value;
and calculating the ratio of the loss value of each user characteristic to the total loss value, and taking the ratio as the loss value ratio of each user characteristic.
In one embodiment, the loss value ratios of the user characteristics are arranged in a descending order to generate a loss value sequence;
accumulating the loss value sequence from the maximum value in sequence to obtain the accumulated value of the loss values;
when the accumulated value is larger than or equal to a preset accumulated threshold value, taking the user characteristic corresponding to the accumulated value as a user loss characteristic;
and acquiring the user loss reason according to the user loss characteristics.
In one embodiment, according to the user churn characteristics, obtaining a WOE code value corresponding to the user churn characteristics;
generating a user loss reason according to a preset loss reason mapping table by using the user loss characteristics of which the WOE coding value exceeds a preset WOE coding threshold value; the churn reason mapping table comprises mapping relations between user churn characteristics and user churn reasons.
In one embodiment, continuous user characteristics in the target user data are subjected to binning processing, and binning corresponding to each continuous user characteristic is obtained;
calculating the WOE code value of each sub-box and the WOE code value of discrete user characteristics in the target user data;
and setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table.
An apparatus for obtaining a reason for user churn, the apparatus comprising:
the first acquisition module is used for acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
the first calculation module is used for calculating the loss value of each user characteristic according to the loss user data through a preset loss value calculation formula; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn;
the second calculation module is used for calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
and the second acquisition module is used for acquiring the user loss reason according to the loss value ratio of each user characteristic.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn;
calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
and acquiring the user loss reason according to the loss value ratio of each user characteristic.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn;
calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
and acquiring the user loss reason according to the loss value ratio of each user characteristic.
According to the method, the device, the computer equipment and the storage medium for acquiring the user loss reason, the loss value of each user characteristic is calculated through a preset loss value calculation formula and loss user data; the attrition user data comprises at least one user characteristic of the attrition user; and calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic, and acquiring the loss reason of each user. Not only can the prediction on whether the user loses be realized, but also the accurate acquisition of the loss reason of the user is realized.
Drawings
FIG. 1 is a diagram of an exemplary embodiment of an application environment for a method for obtaining a cause of user churn;
fig. 2 is a flowchart illustrating a method for obtaining a cause of user churn in one embodiment;
FIG. 3 is a flowchart illustrating a step of obtaining a cause of user churn in one embodiment;
FIG. 4 is a block diagram of an apparatus for obtaining a cause of user churn in one embodiment;
FIG. 5 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The method for acquiring the user churn reason can be applied to the application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The terminal 102 and the server 104 may be respectively and independently used to execute the method for acquiring the user churn reason provided by the present application. The terminal 102 and the server 104 may also be configured to cooperatively execute the method for acquiring the user churn reason provided by the present application. For example, server 104 is used to obtain attrition user data; the attrition user data comprises at least one user characteristic of the attrition user; calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn; calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features; and acquiring the user loss reason according to the loss value ratio of each user characteristic.
The terminal 102 may be, but is not limited to, a device including a data acquisition apparatus, and the server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers.
In one embodiment, as shown in fig. 2, a method for extracting a video key frame is provided, which is described by taking the method as an example applied to the terminal in fig. 1, and includes the following steps:
step 202, acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user.
The lost user data is related data of lost users and comprises not less than one user characteristic; for example, including the user's occupation, age, gender, school calendar, length of the account, etc.
Specifically, firstly, the loss user data needs to be acquired, the acquisition mode of the loss user data is not specifically limited, and the loss user data can be acquired through statistical acquisition and model judgment; the attrition user data includes user characteristics of attrition users, including discrete user characteristics and/or continuous user characteristics. And finally acquiring the loss reason of the user by analyzing the user characteristics of the user.
Step 204, calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn.
Specifically, the churn value calculation formula is a formula for calculating a churn value of each user characteristic of the user according to churn user data; the churn value of a user characteristic is a parameter for characterizing the relationship between the user characteristic of a user as a churn cause and the result of user churn. For a user characteristic of a user, the larger the loss value of the user characteristic is, the stronger the user characteristic plays a role in user loss; the smaller the churn value of the user characteristic, the weaker the user characteristic will have on the user churn.
Step 206, calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values for each user characteristic.
Specifically, the churn value of each user feature can better reflect the role of each user feature on user churn in all user features. Therefore, it is necessary to obtain the loss value ratio of each user feature according to the loss value of each user feature and the total loss value obtained according to the sum of the loss values of all user features. In general, the higher the loss value of a user feature is, the greater the user feature plays a role in user loss; conversely, the lower the churn value duty of a user feature, the less the user feature will contribute to user churn.
And step 208, acquiring the loss reason of the user according to the loss value ratio of each user characteristic.
Specifically, the user characteristics with a large user loss effect can be screened out through the loss value ratio of each user characteristic; the user characteristics with a large user loss effect are analyzed, the relation between the user characteristics and the user loss reasons is searched, and the specific reason of the user loss can be found.
In the method for acquiring the user loss reason, the loss value of each user characteristic is calculated through a preset loss value calculation formula and loss user data; the attrition user data comprises at least one user characteristic of the attrition user; calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the total churn value is the sum of the churn values of the user features; finally, acquiring the loss reason of the user according to the loss value ratio of each user characteristic; the method can convert the user characteristics causing user loss into specific service language and accurately acquire the loss reason of the user.
In one embodiment, the obtaining attrition user data comprises:
inputting target user data into a pre-trained lost user judgment model, and outputting lost user data;
the loss user judgment model is obtained by training according to sample user data and a loss label of a sample user; the sample user data comprises not less than one user characteristic of a sample user; the attrition label is a parameter used to characterize whether the sample user is an attrition user.
Specifically, the target user data includes churned user data and non-churned user data, and the target user data includes user characteristics of churned users and non-churned users. The sample user data and the attrition labels of the sample users are training sets used for training the attrition user judgment model. The loss user judgment model is obtained by training based on sample user data and a loss label of a sample user, and the target user data can be screened and the loss user data can be screened out through the loss user judgment model, so that the loss user data can be analyzed and the loss reason of the loss user can be obtained.
In the embodiment, the loss user judgment model is obtained by training according to the sample user data and the loss label of the sample user as a training set, and the target user data is screened through the loss user judgment model, so that the loss user data can be rapidly screened out, and the efficiency of obtaining the loss reason of the user is improved.
In an embodiment, the calculating, by using a preset churn value calculation formula, a churn value of each user feature includes:
substituting the loss user data into a preset loss value calculation formula to obtain the loss value of each user characteristic;
the loss value calculation formula is as follows:
wherein "! "represents factorial," | "represents the number of elements contained in a set, F represents a set of all user characteristics,
representing the set of features left after the ith feature is removed from the set of F, S represents a subset of F,
representing an attrition user judgment model trained after adding the ith feature to the feature subset S,
an attrition user judgment model trained based on the user feature subset S is represented.
Specifically, the churn value calculation formula is a calculation formula for calculating a churn value of each user characteristic; through the loss value calculation formula, the loss value of each user characteristic can be quickly obtained, so that the loss value ratio can be further obtained according to the loss value of the user. For a user characteristic of a user, the larger the loss value of the user characteristic is, the stronger the user characteristic plays a role in user loss; the smaller the churn value of the user characteristic, the weaker the user characteristic will have on the user churn.
In this embodiment, through the loss value calculation formula, the loss value corresponding to each user characteristic can be quickly obtained, and then the loss reason of the lost user can be quickly obtained according to the loss value of each user characteristic, so that the accuracy of obtaining the loss reason of the lost user is improved.
In an embodiment, the calculating the churn value ratio of each user characteristic according to the churn value of each user characteristic includes:
replacing data smaller than 0 in the loss values of the user characteristics with 0, and calculating the sum of the loss values of the user characteristics to serve as a total loss value;
and calculating the ratio of the loss value of each user characteristic to the total loss value, and taking the ratio as the loss value ratio of each user characteristic.
Specifically, after the churn value of each user characteristic is obtained, the churn value of some user characteristics may be a negative value. The sum of the user characteristics is acquired according to the loss value of each user characteristic to serve as a total loss value, and the loss value ratio of each user characteristic is further acquired according to the loss value and the total loss value of each user characteristic, so that 0 is adopted to replace the value smaller than 0 in the loss values of the user characteristics. And calculating the total loss value and the loss value ratio of each user characteristic according to the loss value of the replaced user characteristic.
The calculation formula of the loss value ratio is as follows:
wherein the content of the first and second substances,
an attrition value representing a jth user characteristic of an ith attrition user;
to representThe loss value of the jth user characteristic of the ith lost user accounts for the ratio; n is the number of user features of the attrition users.
In this embodiment, the churn value of a user profile may be a positive value or a negative value, and a negative churn value indicates that the user profile does not cause churn. Therefore, in order to eliminate the influence of the user characteristics which do not cause user loss, a negative value in the loss value is replaced by 0 in the calculation process, and then a characteristic with a larger proportion is searched in a mode of calculating the proportion of the loss value from the characteristics with the positive value of the loss value, so that the important user characteristics are found.
In an embodiment, the obtaining the user churn reason according to the churn value ratio of each user feature includes:
performing descending order arrangement on the loss value ratios of the user characteristics to generate a loss value sequence;
accumulating the loss value sequence from the maximum value in sequence to obtain the accumulated value of the loss values;
when the accumulated value is larger than or equal to a preset accumulated threshold value, taking the user characteristic corresponding to the accumulated value as a user loss characteristic;
and acquiring the user loss reason according to the user loss characteristics.
Specifically, among the user characteristics of each attrition user, only a small portion of the user characteristics are strongly correlated with the attrition user. User churn features that are not strongly correlated with churn users need to be removed. Therefore, after the loss values of all the user characteristics of the lost users are obtained, the loss value ratios of all the user characteristics are arranged in a descending order to generate a loss value sequence; and accumulating the loss value sequence from the maximum value in sequence to obtain the accumulated value of the loss values. And obtaining the user characteristics with the top rank as the user loss characteristics according to the accumulated values and the accumulated threshold values, and obtaining the user loss reasons according to the user loss characteristics.
In this embodiment, some of the user characteristics with the positive loss values of the user characteristics are relatively large (for example, 5), some of the user characteristics are relatively small (for example, 0.005), the user characteristics with the relatively large loss values have a relatively large effect on the user loss, and similarly, the user characteristics with the small loss values have a relatively small effect on the user loss. Therefore, by calculating the churn value fraction, it is possible to find several user characteristics that have the greatest effect on user churn from all the user characteristics.
As shown in fig. 3, the obtaining the user churn reason according to the user churn feature includes:
step 302, obtaining a WOE code value corresponding to the user loss characteristic according to the user loss characteristic;
specifically, for continuous user churn features, binning processing is required to be performed, binning corresponding to each continuous user churn feature is obtained, and a WOE (Evidence Weight) coding value of each bin is calculated; for discrete user churn characteristics, the WOE code values of the discrete user churn characteristics can be calculated without binning.
Step 304, generating a user loss reason according to a preset loss reason mapping table by using the user loss characteristics of which the WOE coding value exceeds a preset WOE coding threshold value; the churn reason mapping table comprises mapping relations between user churn characteristics and user churn reasons.
Specifically, the WOE code value is obtained by the following formula:
wherein the content of the first and second substances,
is the WOE encoded value of the ith packet,
is the number of attrition users in the current interval,
is the number of users in the current interval that are not churned,
indicating the number of attrition users in all samples under the user profile,
indicating the number of users in all samples for which churn did not occur under this feature.
Acquiring a WOE (word on element) code value corresponding to user loss characteristics, judging whether the WOE code value corresponding to the user loss characteristics exceeds a preset WOE code threshold value, and if the WOE code value exceeds the preset WOE code threshold value, generating corresponding user loss reasons through a loss reason mapping table containing mapping relations between the user loss characteristics and the user loss reasons.
In the embodiment, a WOE code value corresponding to the user churn characteristic is generated according to the user churn characteristic; and generating a user loss reason according to a preset loss reason mapping table by using the user loss characteristics of which the WOE code value exceeds a preset WOE code threshold value. The method can convert the user characteristics causing user loss into specific service language and accurately acquire the loss reason of the user.
In an embodiment, the obtaining, according to the user churn feature, a WOE code value corresponding to the user churn feature previously further includes:
performing binning processing on continuous user characteristics in the target user data to obtain bins corresponding to the continuous user characteristics;
calculating the WOE code value of each sub-box and the WOE code value of discrete user characteristics in the target user data;
and setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table.
Specifically, the user features in the target user data include continuous user features and discrete user features, for the continuous user features, binning processing needs to be performed to obtain bins corresponding to each continuous user feature, and a WOE code value of each bin is calculated; for discrete user characteristics, the WOE code value of the discrete user characteristics is directly calculated through a formula. And finally, setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table. The reason for user churn is described in the business language.
For example, for a discrete user characteristic of "user account opening time", the relationship between the WOE interval, the WOE code value and the service language indicating the reason for user churn is shown in table 1:
TABLE 1
The WOE interval in the table 1 is customized according to needs, and the WOE code value is calculated and obtained through an obtaining formula of the WOE code value; the value of the WOE coding threshold in table 1 is 0.1, and for a WOE interval in which the WOE coding value is greater than 0.1, a corresponding user loss reason described by a service language is set. As shown in table 1, when the account opening duration of the user is less than or equal to 188 days, the user churn reason of the churn user is a new account opening; when the account opening time of the user is more than 188 days and less than or equal to 714 days, the user churn reason of the churn user is that the account opening time is short.
In this embodiment, continuous user features in the target user data are subjected to binning processing to obtain bins corresponding to each continuous user feature, and a WOE code value of each bin and a WOE code value of discrete user features in the target user data are calculated. And finally, setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table. Therefore, the user loss reason can be acquired according to the user loss characteristic and the user loss reason mapping table.
It should be understood that although the various steps in the flow charts of fig. 2-3 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2-3 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed in turn or alternately with other steps or at least some of the other steps.
In one embodiment, as shown in fig. 4, an apparatus for obtaining a cause of user churn is provided, including: a first obtaining module 401, a first calculating module 402, a second calculating module 403, and a second obtaining module 404, wherein:
a first obtaining module 401, configured to obtain churn user data; the attrition user data comprises at least one user characteristic of the attrition user.
A first calculating module 402, configured to calculate a churn value of each user feature according to the churn user data through a preset churn value calculation formula; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn.
A second calculating module 403, configured to calculate a ratio of loss values of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values for each user characteristic.
A second obtaining module 404, configured to obtain a user churn reason according to the churn value ratio of each user characteristic.
In an embodiment, the first obtaining module 401 is further configured to input target user data into a pre-trained attrition user determination model, and output attrition user data; the loss user judgment model is obtained by training according to sample user data and a loss label of a sample user; the sample user data comprises not less than one user characteristic of a sample user; the attrition label is a parameter used to characterize whether the sample user is an attrition user.
In an embodiment, the first calculating module 402 is further configured to substitute the churn user data into a preset churn value calculation formula to obtain a churn value of each user feature;
the loss value calculation formula is as follows:
wherein "! "represents factorial," | "represents the number of elements contained in a set, F represents a set of all user characteristics,
representing the set of features left after the ith feature is removed from the set of F, S represents a subset of F,
representing an attrition user judgment model trained after adding the ith feature to the feature subset S,
an attrition user judgment model trained based on the user feature subset S is represented.
In one embodiment, the second calculating module 403 is further configured to replace data smaller than 0 in the churn value of each user feature with 0, and calculate a sum of the churn values of each user feature as a total churn value; and calculating the ratio of the loss value of each user characteristic to the total loss value, and taking the ratio as the loss value ratio of each user characteristic.
In an embodiment, the second obtaining module 404 is further configured to perform descending order arrangement on the attrition value ratios of the user features, so as to generate an attrition value sequence; accumulating the loss value sequence from the maximum value in sequence to obtain the accumulated value of the loss values; when the accumulated value is larger than or equal to a preset accumulated threshold value, taking the user characteristic corresponding to the accumulated value as a user loss characteristic; and acquiring the user loss reason according to the user loss characteristics.
In an embodiment, the second obtaining module 404 is further configured to obtain, according to the user churn feature, a WOE code value corresponding to the user churn feature; generating a user loss reason according to a preset loss reason mapping table by using the user loss characteristics of which the WOE coding value exceeds a preset WOE coding threshold value; the churn reason mapping table comprises mapping relations between user churn characteristics and user churn reasons.
In an embodiment, the second obtaining module 404 is further configured to perform binning processing on consecutive user features in the target user data, and obtain bins corresponding to each consecutive user feature; calculating the WOE code value of each sub-box and the WOE code value of discrete user characteristics in the target user data; and setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table.
In the device for acquiring the user loss reason, the loss value of each user characteristic is calculated through a preset loss value calculation formula and loss user data; the attrition user data comprises at least one user characteristic of the attrition user; calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the total churn value is the sum of the churn values of the user features; finally, acquiring the loss reason of the user according to the loss value ratio of each user characteristic; the method can convert the user characteristics causing user loss into specific service language and accurately acquire the loss reason of the user.
For specific limitations of the apparatus for acquiring the user churn cause, reference may be made to the above limitations on the method for acquiring the user churn cause, which are not described herein again. All or part of the modules in the device for acquiring the user churn cause can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 5. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method for acquiring a cause of user churn.
Those skilled in the art will appreciate that the architecture shown in fig. 5 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn;
calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
and acquiring the user loss reason according to the loss value ratio of each user characteristic.
In one embodiment, the processor, when executing the computer program, further performs the steps of: inputting target user data into a pre-trained lost user judgment model, and outputting lost user data; the loss user judgment model is obtained by training according to sample user data and a loss label of a sample user; the sample user data comprises not less than one user characteristic of a sample user; the attrition label is a parameter used to characterize whether the sample user is an attrition user.
In one embodiment, the processor, when executing the computer program, further performs the steps of: substituting the loss user data into a preset loss value calculation formula to obtain the loss value of each user characteristic;
the loss value calculation formula is as follows:
wherein "! "represents factorial," | "represents the number of elements contained in a set, F represents a set of all user characteristics,
representing the set of features left after the ith feature is removed from the set of F, S represents a subset of F,
representing an attrition user judgment model trained after adding the ith feature to the feature subset S,
an attrition user judgment model trained based on the user feature subset S is represented.
In one embodiment, the processor, when executing the computer program, further performs the steps of: replacing data smaller than 0 in the loss values of the user characteristics with 0, and calculating the sum of the loss values of the user characteristics to serve as a total loss value; and calculating the ratio of the loss value of each user characteristic to the total loss value, and taking the ratio as the loss value ratio of each user characteristic.
In one embodiment, the processor, when executing the computer program, further performs the steps of: performing descending order arrangement on the loss value ratios of the user characteristics to generate a loss value sequence; accumulating the loss value sequence from the maximum value in sequence to obtain the accumulated value of the loss values; when the accumulated value is larger than or equal to a preset accumulated threshold value, taking the user characteristic corresponding to the accumulated value as a user loss characteristic; and acquiring the user loss reason according to the user loss characteristics.
In one embodiment, the processor, when executing the computer program, further performs the steps of: acquiring a WOE (word on edge) coding value corresponding to the user loss characteristic according to the user loss characteristic; generating a user loss reason according to a preset loss reason mapping table by using the user loss characteristics of which the WOE coding value exceeds a preset WOE coding threshold value; the churn reason mapping table comprises mapping relations between user churn characteristics and user churn reasons.
In one embodiment, the processor, when executing the computer program, further performs the steps of: performing binning processing on continuous user characteristics in the target user data to obtain bins corresponding to the continuous user characteristics; calculating the WOE code value of each sub-box and the WOE code value of discrete user characteristics in the target user data; and setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table.
The computer equipment calculates the loss value of each user characteristic through a preset loss value calculation formula and loss user data; the attrition user data comprises at least one user characteristic of the attrition user; calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the total churn value is the sum of the churn values of the user features; finally, acquiring the loss reason of the user according to the loss value ratio of each user characteristic; the method can convert the user characteristics causing user loss into specific service language and accurately acquire the loss reason of the user.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn;
calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
and acquiring the user loss reason according to the loss value ratio of each user characteristic.
In one embodiment, the computer program when executed by the processor further performs the steps of: inputting target user data into a pre-trained lost user judgment model, and outputting lost user data; the loss user judgment model is obtained by training according to sample user data and a loss label of a sample user; the sample user data comprises not less than one user characteristic of a sample user; the attrition label is a parameter used to characterize whether the sample user is an attrition user.
In one embodiment, the computer program when executed by the processor further performs the steps of: substituting the loss user data into a preset loss value calculation formula to obtain the loss value of each user characteristic;
the loss value calculation formula is as follows:
wherein "! "represents factorial," | "represents the number of elements contained in a set, F represents a set of all user characteristics,
representing the set of features left after the ith feature is removed from the set of F, S represents a subset of F,
representing an attrition user judgment model trained after adding the ith feature to the feature subset S,
an attrition user judgment model trained based on the user feature subset S is represented.
In one embodiment, the computer program when executed by the processor further performs the steps of: replacing data smaller than 0 in the loss values of the user characteristics with 0, and calculating the sum of the loss values of the user characteristics to serve as a total loss value; and calculating the ratio of the loss value of each user characteristic to the total loss value, and taking the ratio as the loss value ratio of each user characteristic.
In one embodiment, the computer program when executed by the processor further performs the steps of: performing descending order arrangement on the loss value ratios of the user characteristics to generate a loss value sequence; accumulating the loss value sequence from the maximum value in sequence to obtain the accumulated value of the loss values; when the accumulated value is larger than or equal to a preset accumulated threshold value, taking the user characteristic corresponding to the accumulated value as a user loss characteristic; and acquiring the user loss reason according to the user loss characteristics.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a WOE (word on edge) coding value corresponding to the user loss characteristic according to the user loss characteristic; generating a user loss reason according to a preset loss reason mapping table by using the user loss characteristics of which the WOE coding value exceeds a preset WOE coding threshold value; the churn reason mapping table comprises mapping relations between user churn characteristics and user churn reasons.
In one embodiment, the computer program when executed by the processor further performs the steps of: performing binning processing on continuous user characteristics in the target user data to obtain bins corresponding to the continuous user characteristics; calculating the WOE code value of each sub-box and the WOE code value of discrete user characteristics in the target user data; and setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table.
The storage medium calculates the loss value of each user characteristic through a preset loss value calculation formula and loss user data; the attrition user data comprises at least one user characteristic of the attrition user; calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the total churn value is the sum of the churn values of the user features; finally, acquiring the loss reason of the user according to the loss value ratio of each user characteristic; the method can convert the user characteristics causing user loss into specific service language and accurately acquire the loss reason of the user.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (9)
1. A method for acquiring a user churn reason is characterized by comprising the following steps:
acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
calculating the loss value of each user characteristic through a preset loss value calculation formula according to the loss user data; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn;
calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
acquiring a user loss reason according to the loss value ratio of each user characteristic;
the calculating the loss value of each user characteristic through a preset loss value calculation formula comprises:
substituting the loss user data into a preset loss value calculation formula to obtain the loss value of each user characteristic;
the loss value calculation formula is as follows:
wherein "! "represents factorial," | "represents the number of elements contained in a set, F represents a set of all user characteristics,
representing the set of features left after the ith feature is removed from the set of F, S represents a subset of F,
representing an attrition user judgment model trained after adding the ith feature to S,
representing the union of S and the ith feature;
representing an attrition user judgment model based on S training.
2. The method of claim 1, wherein obtaining attrition user data comprises:
inputting target user data into a pre-trained lost user judgment model, and outputting lost user data;
the loss user judgment model is obtained by training according to sample user data and a loss label of a sample user; the sample user data comprises not less than one user characteristic of a sample user; the attrition label is a parameter used to characterize whether the sample user is an attrition user.
3. The method of claim 1, wherein calculating the churn value ratio for each user profile based on the churn value for each user profile comprises:
replacing data smaller than 0 in the loss values of the user characteristics with 0, and calculating the sum of the loss values of the user characteristics to serve as a total loss value;
and calculating the ratio of the loss value of each user characteristic to the total loss value, and taking the ratio as the loss value ratio of each user characteristic.
4. The method according to claim 2, wherein the obtaining the user churn reason according to the churn value ratio of each user feature comprises:
performing descending order arrangement on the loss value ratios of the user characteristics to generate a loss value sequence;
accumulating the loss value sequence from the maximum value in sequence to obtain the accumulated value of the loss values;
when the accumulated value is larger than or equal to a preset accumulated threshold value, taking the user characteristic corresponding to the accumulated value as a user loss characteristic;
and acquiring the user loss reason according to the user loss characteristics.
5. The method according to claim 4, wherein the obtaining a user churn cause according to the user churn feature comprises:
acquiring a WOE (word on edge) coding value corresponding to the user loss characteristic according to the user loss characteristic;
generating a user loss reason according to a preset loss reason mapping table by using the user loss characteristics of which the WOE coding value exceeds a preset WOE coding threshold value; the churn reason mapping table comprises mapping relations between user churn characteristics and user churn reasons.
6. The method of claim 5, wherein obtaining the WOE code value corresponding to the user churn feature according to the user churn feature further comprises:
performing binning processing on continuous user characteristics in the target user data to obtain bins corresponding to the continuous user characteristics;
calculating the WOE code value of each sub-box and the WOE code value of discrete user characteristics in the target user data;
and setting a corresponding user loss reason for the user characteristics of which the WOE code value is greater than the preset WOE code threshold value, and generating a user loss reason mapping table.
7. An apparatus for obtaining a reason for user churn, the apparatus comprising:
the first acquisition module is used for acquiring lost user data; the attrition user data comprises at least one user characteristic of the attrition user;
the first calculation module is used for calculating the loss value of each user characteristic according to the loss user data through a preset loss value calculation formula; the churn value is a parameter used to quantify the relationship between the user characteristics and user churn; the calculating the loss value of each user characteristic through a preset loss value calculation formula comprises:
substituting the loss user data into a preset loss value calculation formula to obtain the loss value of each user characteristic;
the loss value calculation formula is as follows:
wherein "! "represents factorial," | "represents the number of elements contained in a set, F represents a set of all user characteristics,
representing the set of features left after the ith feature is removed from the set of F, S represents a subset of F,
representing an attrition user judgment model trained after adding the ith feature to S,
representing the union of S and the ith feature;
representing an attrition user judgment model based on S training;
the second calculation module is used for calculating the loss value ratio of each user characteristic according to the loss value of each user characteristic; the loss value proportion is the proportion of the loss values of all user characteristics in the total loss value; the total churn value is the sum of the churn values of the user features;
and the second acquisition module is used for acquiring the user loss reason according to the loss value ratio of each user characteristic.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 6.
9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110293420.3A CN112686718B (en) | 2021-03-19 | 2021-03-19 | Method and device for acquiring user loss reason, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110293420.3A CN112686718B (en) | 2021-03-19 | 2021-03-19 | Method and device for acquiring user loss reason, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112686718A CN112686718A (en) | 2021-04-20 |
| CN112686718B true CN112686718B (en) | 2021-06-29 |
Family
ID=75455665
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110293420.3A Active CN112686718B (en) | 2021-03-19 | 2021-03-19 | Method and device for acquiring user loss reason, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112686718B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105099731A (en) * | 2014-04-23 | 2015-11-25 | 腾讯科技（深圳）有限公司 | Method and system for finding churn factor for user churn of network application |
| CN108510096A (en) * | 2017-02-24 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | Trade company's attrition prediction method, apparatus, equipment and storage medium |
| AU2019101198A4 (en) * | 2019-10-03 | 2020-01-16 | Gu, Jingchao MR | A statistical analysis method of mobile telecom data driven user loss prediction |
| CN110852780A (en) * | 2019-10-08 | 2020-02-28 | 百度在线网络技术（北京）有限公司 | Data analysis method, device, equipment and computer storage medium |
| CN111652661A (en) * | 2020-08-04 | 2020-09-11 | 索信达(北京)数据技术有限公司 | Mobile phone client user loss early warning processing method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105335430B (en) * | 2014-08-08 | 2019-09-17 | 腾讯科技（深圳）有限公司 | A kind of customer churn determining method of path and device based on map scene |
-
2021
- 2021-03-19 CN CN202110293420.3A patent/CN112686718B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105099731A (en) * | 2014-04-23 | 2015-11-25 | 腾讯科技（深圳）有限公司 | Method and system for finding churn factor for user churn of network application |
| CN108510096A (en) * | 2017-02-24 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | Trade company's attrition prediction method, apparatus, equipment and storage medium |
| AU2019101198A4 (en) * | 2019-10-03 | 2020-01-16 | Gu, Jingchao MR | A statistical analysis method of mobile telecom data driven user loss prediction |
| CN110852780A (en) * | 2019-10-08 | 2020-02-28 | 百度在线网络技术（北京）有限公司 | Data analysis method, device, equipment and computer storage medium |
| CN111652661A (en) * | 2020-08-04 | 2020-09-11 | 索信达(北京)数据技术有限公司 | Mobile phone client user loss early warning processing method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112686718A (en) | 2021-04-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109598095B (en) | Method and device for establishing scoring card model, computer equipment and storage medium | |
| CN109829629B (en) | Risk analysis report generation method, apparatus, computer device and storage medium | |
| CN111222305B (en) | Information structuring method and device | |
| US20200065710A1 (en) | Normalizing text attributes for machine learning models | |
| CN108520041B (en) | Industry classification method and system of text, computer equipment and storage medium | |
| CN110263326B (en) | User behavior prediction method, prediction device, storage medium and terminal equipment | |
| CN113240510B (en) | Abnormal user prediction method, device, equipment and storage medium | |
| CN112395500A (en) | Content data recommendation method and device, computer equipment and storage medium | |
| CN110888911A (en) | Sample data processing method and device, computer equipment and storage medium | |
| CN111190946A (en) | Report generation method and device, computer equipment and storage medium | |
| CN113268403A (en) | Time series analysis and prediction method, device, equipment and storage medium | |
| CN111160959B (en) | User click conversion prediction method and device | |
| CN113254620B (en) | Response method, device and equipment based on graph neural network and storage medium | |
| CN110602207A (en) | Method, device, server and storage medium for predicting push information based on off-network | |
| CN111599382B (en) | Voice analysis method, device, computer equipment and storage medium | |
| CN114203200A (en) | Voice quality inspection method and device, computer equipment and storage medium | |
| CN112686718B (en) | Method and device for acquiring user loss reason, computer equipment and storage medium | |
| CN111104422A (en) | Training method, device, equipment and storage medium of data recommendation model | |
| US20230214451A1 (en) | System and method for finding data enrichments for datasets | |
| CN113946363A (en) | Method and device for executing and configuring service data, computer equipment and storage medium | |
| CN113438374B (en) | Intelligent outbound call processing method, device, equipment and storage medium | |
| CN111078984B (en) | Network model issuing method, device, computer equipment and storage medium | |
| CN114141235A (en) | Voice corpus generation method and device, computer equipment and storage medium | |
| CN113627514A (en) | Data processing method and device of knowledge graph, electronic equipment and storage medium | |
| CN111368044A (en) | Intelligent question answering method and device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |