CN115330447A - User investment prediction training method, device and computer readable storage medium - Google Patents
User investment prediction training method, device and computer readable storage medium Download PDFInfo
- Publication number
- CN115330447A CN115330447A CN202210959399.0A CN202210959399A CN115330447A CN 115330447 A CN115330447 A CN 115330447A CN 202210959399 A CN202210959399 A CN 202210959399A CN 115330447 A CN115330447 A CN 115330447A
- Authority
- CN
- China
- Prior art keywords
- prediction
- users
- user
- group
- funding
- 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.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 106
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000012216 screening Methods 0.000 claims abstract description 19
- 238000006243 chemical reaction Methods 0.000 claims description 71
- 238000000546 chi-square test Methods 0.000 claims description 22
- 239000011159 matrix material Substances 0.000 claims description 18
- 238000012795 verification Methods 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 6
- 230000006399 behavior Effects 0.000 description 35
- 230000014759 maintenance of location Effects 0.000 description 10
- 238000012360 testing method Methods 0.000 description 7
- 238000001358 Pearson's chi-squared test Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000007477 logistic regression Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000007637 random forest analysis Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012797 qualification Methods 0.000 description 3
- 230000004927 fusion Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000033228 biological regulation Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
Abstract
The embodiment of the invention relates to the technical field of Internet, and discloses a user investment prediction training method, a device and a computer readable storage medium, wherein the method comprises the following steps: acquiring user prediction characteristics; wherein the user prediction features comprise user features in public domain data and user features in private domain data; the user prediction features determine the features to be trained related to the investment prediction through a preset screening rule; training through a training sample containing the characteristics to be trained to obtain a target investment prediction model, and applying the technical scheme of the invention, all user prediction characteristics contained in public and private domain data can be integrated; the reserve can be predicted more accurately.
Description
Technical Field
The embodiment of the invention relates to the technical field of Internet, in particular to a user investment prediction training method and device and a computer readable storage medium.
Background
With the development of the automobile industry in the digital marketing technology, potential consumers can be effectively mined by utilizing the internet, the computer communication technology and the digital interactive media.
Under the digital marketing technology of the automobile industry, potential consumers are mined mainly through the intension of paying (leaving data) of possible clues of public domain data of user groups. However, the population of potential consumers obtained under the public domain data, and the willingness of that population to fund, do not provide accurate potential consumers and a prediction of the fund.
Disclosure of Invention
In view of the foregoing problems, embodiments of the present invention provide a user funding prediction training method, device and computer-readable storage medium, which are used to solve the problem that potential consumers cannot be accurately mined due to the possible funding intention of a user group provided by public domain data in the prior art.
According to an aspect of an embodiment of the present invention, there is provided a user funding prediction training method, including:
acquiring user prediction characteristics; wherein the user prediction features comprise user features in public domain data and user features in private domain data;
the user prediction features determine the features to be trained related to the investment prediction through a preset screening rule;
and training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
In an optional manner, the training by using the training sample including the feature to be trained to obtain the target funding prediction model further includes:
training at least two investment prediction models based on the training sample containing the characteristics to be trained;
obtaining the capital-reserving prediction accuracy rate and the mixed matrix index output by at least two capital-reserving prediction models;
and determining an optimal remaining resource prediction model suitable for remaining resource prediction in at least two remaining resource prediction models as a target remaining resource prediction model according to the remaining resource prediction accuracy and the mixed matrix index.
In an optional manner, the training by using the training sample including the feature to be trained to obtain the target funding prediction model further includes:
iterating the features to be trained in the training samples;
and when the accuracy of the reserve prediction output by the target reserve prediction model is higher than a preset value, taking the feature to be trained in the training sample as a reserve prediction feature.
In an optional manner, after the training is performed by using the training sample including the feature to be trained to obtain the target funding prediction model, the method further includes:
popularizing aiming at a first group of users and a second group of users, and determining the refunding conversion rate of the first group of users and the refunding conversion rate of the second group of users; the first group of users are users with a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users with the first number, randomly selected from the promoted users.
In an optional manner, after determining the funding conversion rate of the first group of users and the funding conversion rate of the second group of users for the first group of users and the second group of users, the method further includes:
and if the capital-remaining conversion rate of the first group of users is higher than the capital-remaining conversion rate of the second group of users, or the capital-remaining conversion rate of the first group of users is higher than a third preset value, taking the target capital-remaining prediction model as a final application capital-remaining prediction model.
In an optional manner, after determining the funding conversion rate of the first group of users and the funding conversion rate of the second group of users for the first group of users and the second group of users, the method further includes:
and if the capital-remaining conversion rate of the first group of users is lower than or equal to the capital-remaining conversion rate of the second group of users, or the capital-remaining conversion rate of the first group of users is lower than or equal to a third preset value, optimizing the target capital-remaining prediction model through the characteristics to be trained.
In an optional manner, the preset screening rule is chi-square test, and the user predicted feature determines a feature to be trained related to the investment prediction through the preset screening rule, further including:
obtaining a statistic of the user prediction characteristic and a degree of freedom of the statistic;
and determining the characteristic to be trained related to the vestigial prediction according to the statistic value, the degree of freedom and the linear correlation of the Pearson correlation coefficient.
According to another aspect of the embodiments of the present invention, there is provided a user funding prediction apparatus, including:
the acquisition module acquires user prediction characteristics; wherein the user prediction features comprise user features in public domain data and user features in private domain data;
the checking module is used for determining the characteristics to be trained related to the reserve resources prediction by the user prediction characteristics through a preset screening rule;
and the model module is used for training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
In an optional manner, the user funding prediction apparatus further includes:
the verification module is used for popularizing aiming at a first group of users and a second group of users and determining the refund conversion rate of the first group of users and the refund conversion rate of the second group of users; the first group of users are users with a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users with the first number, randomly selected from the promoted users.
According to another aspect of the embodiments of the present invention, there is provided a computer-readable storage medium, in which at least one executable instruction is stored, and when the executable instruction is executed on a user investment prediction device, the user investment prediction device executes the operations of the user investment prediction training method described in the summary of the present invention.
According to the embodiment of the invention, all user prediction characteristics contained in public and private domain data can be integrated by acquiring the user prediction characteristics comprising the user characteristics in the public domain data and the user characteristics in the private domain data; the user prediction features can determine the features to be trained related to the investment prediction through a preset screening rule; training through a training sample containing the characteristics to be trained to obtain a target investment prediction model; the reserve can be predicted more accurately.
The foregoing description is only an overview of the technical solutions of the embodiments of the present invention, and in order that the technical solutions of the embodiments of the present invention can be clearly understood, the embodiments of the present invention can be implemented according to the content of the description, and the above and other objects, features, and advantages of the embodiments of the present invention can be more clearly understood, the detailed description of the present invention is provided below.
Drawings
The drawings are only for purposes of illustrating embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 is a flow chart illustrating a first embodiment of a user funding prediction training method provided by the present invention;
FIG. 2a is a schematic flow chart illustrating a second embodiment of the user funding prediction training method provided by the present invention;
FIG. 2b is a schematic diagram illustrating an example of querying a user based on location information data in the user funding prediction training method provided by the present invention;
FIG. 2c is a schematic flow chart showing the determination of a target investment prediction model in the user investment prediction training method provided by the invention;
FIG. 2d shows a schematic flow chart of selecting an optimal investment reservation prediction model suitable for investment reservation prediction as a target investment reservation prediction model in the user investment reservation prediction training method provided by the invention;
FIG. 3 is a flowchart illustrating a third embodiment of the user funding prediction training method provided by the present invention;
fig. 4 shows a schematic structural diagram of an embodiment of the user funding prediction device provided by the present invention.
Detailed Description
Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be embodied in various forms and should not be limited to the embodiments set forth herein.
Fig. 1 is a flowchart of a first embodiment of a user funding prediction training method according to the present invention, which is performed by a user funding prediction training method apparatus. As shown in fig. 1, the method comprises the steps of:
step 110: acquiring user prediction characteristics; wherein the user predicted characteristics include user characteristics in public domain data and user characteristics in private domain data.
The public domain data is data of a public domain flow platform obtained by directly entering a mature platform for automobile enterprises (vehicle enterprises for short) to realize flow conversion.
Private domain data is data which is owned by the vehicle enterprises in the brands, does not need to pay, can be repeatedly utilized and can obtain user flow at any time; the user characteristics can be user characteristics obtained by a user in an automobile brand private domain, and also can be user characteristics obtained in a new energy automobile brand private domain, wherein the private domain can comprise channels such as a brand official website, a brand APP (Application, mobile phone Application), a brand applet and the like.
Step 120: and determining the characteristics to be trained related to the reserve prediction by the user prediction characteristics through a preset screening rule.
The prediction screening rule can screen out the user prediction characteristics related to the investment reservation side from all the user prediction characteristics, and the user prediction characteristics are used as the characteristics to be trained.
Step 130: and training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
Wherein, the training sample may be a sample only containing the feature to be trained.
In some embodiments, the training samples may also be positive and negative samples, that is, the training samples may include positive samples and negative samples, where the positive samples include the features to be trained, and the features to be trained may include features to be trained having high correlation of funded results through a preset screening rule; negative examples may not include the feature to be trained.
And training a plurality of preset withholding prediction models based on the training samples, and obtaining the target withholding prediction model from the preset withholding prediction models.
According to the embodiment of the invention, all the user prediction characteristics contained in the public and private domain data can be integrated by acquiring the user prediction characteristics comprising the user characteristics in the public domain data and the user characteristics in the private domain data; the user prediction features can determine the features to be trained related to the investment prediction through a preset screening rule; training through a training sample containing the characteristics to be trained to obtain a target investment prediction model; all user prediction characteristics contained in the public and private domain data can be integrated; the reserve can be predicted more accurately.
FIG. 2a is a flow chart of another embodiment of the user funding prediction training method of the present invention, which is performed by a user funding prediction device. As shown in fig. 2a, the method comprises the steps of:
step 210: acquiring user prediction characteristics; wherein the user prediction features comprise user features in public domain data and user features in private domain data.
The public domain data is data of a public domain flow platform obtained by directly entering a mature platform for automobile enterprises (vehicle enterprises for short) to realize flow conversion. The private domain data can be user characteristics obtained by a user in an automobile brand private domain, and can also be user characteristics obtained in a new energy automobile brand private domain, wherein the private domain can comprise a brand official website, a brand APP (Application, mobile phone Application), a brand applet and the like.
The user characteristics include user attributes and user behavior.
The user attributes may include user basic information, vehicle purchasing intention and/or congestion measurement system values. The basic information of the user can comprise gender, age bracket, frequent residence, family resident population, income grade, whether a car exists or not and the like; the buying intent may include vehicle type preferences, energy preferences and/or buying budget; the user metric system values may include an activity force value, an influence force value, a consumption force value, and/or a total user metric value.
The user behavior of the public domain data includes advertisement exposure and advertisement click, and may include a media platform, an advertisement type, a content type, and N days or YTD exposure times (i.e., exposure times from the beginning of the year to the present).
The user behaviors of the private domain data comprise behaviors on an official website, an applet and/or an application program, and the official website behaviors can comprise browsing days in N days and video playing numbers and times in N days; the small program behavior comprises whether a mobile phone number is authorized or not, the login days and times in N days, and the video playing numbers and times in N days; the application behavior comprises registration, login days and login times in N days, video playing times and video playing times in N days, browsing post number in N days and comment number.
The user behavior is not limited to the specific behaviors listed above, but also includes other behaviors that may trigger user behavior, possibly resulting in user funding.
It should be understood that when user-related data (e.g., user attribute data, user behavior data, user geographic location, etc., where the data type is adapted according to the scenario content) is referred to, it is obtained after permission or approval from the user; that is, when the present application is applied to a specific product or technology, user permission needs to be obtained to achieve the acquisition and processing of the relevant data, and the processing of the relevant data needs to comply with relevant laws and regulations and regulatory standards of relevant countries and regions.
For example, when the current geographic location of the user needs to be obtained, a location obtaining prompt may be displayed in the terminal of the user, and after a confirmation operation of the user for the location obtaining prompt is received, the terminal may obtain the current geographic location of the user, as shown in fig. 2b, it is a process of querying the user based on the location information data, where "whether to allow obtaining the current location information" in the location obtaining prompt determines whether to obtain the current geographic location of the user according to "approval" or "denial" clicked by the user.
Step 220: and the user prediction features are subjected to chi-square test, and the features to be trained related to the investment prediction are determined.
That is, the predetermined filtering rule may be a Chi-Squared Test (Chi-Squared Test) or a Test using other statistical classification inference methods.
And analyzing the characteristic correlation of the user prediction characteristic through chi-square test to determine the characteristic to be trained.
Among them, chi-Squared Test (Chi-Squared Test) subjects the distribution of statistics to a hypothesis Test that approximates a Chi-Squared distribution when a null hypothesis holds. The chi-squared test may refer to the pearson chi-squared test, wherein the chi-squared test is the degree of deviation between the actual observed value and the theoretical inferred value of the statistical sample.
In step 220, the method comprises:
and obtaining a statistic value of the user prediction characteristic and the degree of freedom of the statistic value.
Wherein the test is carried out by the Chi-Square test, in particular the Pearson Chi-Square test. The statistical value is obtained by dividing the difference between the observed value and the theoretical value of the user prediction characteristic by the theoretical value after squaring the difference, and then summing the values. The degree of freedom can be obtained by statistical values.
And determining the characteristic to be trained according to the statistical value, the degree of freedom and the linear correlation of the Pearson correlation coefficient.
The feature to be trained can be determined according to a chi-square distribution critical value and a statistic value of the degree of freedom and the square of the Pearson correlation coefficient r.
Step 230: and training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
Wherein, the training is performed through the training sample containing the feature to be trained to obtain the target investment prediction model, as shown in fig. 2c, the method comprises the following steps:
step 2301: and training at least two investment prediction models based on the training samples containing the features to be trained.
The at least two surviving prediction models may include at least two of a Logistic Regression model (Logistic Regression), a Random Forest Regression model (Random Forest), a Gradient Boosting Tree model (GBDT), and/or a Support Vector Machine (SVM), and may include other ensemble learning models in addition to the above listed models.
Step 2302: and acquiring the remaining prediction accuracy and the mixed matrix index output by at least two remaining prediction models.
Step 2303: and determining an optimal reserve prediction model suitable for reserve prediction in at least two reserve prediction models as a target reserve prediction model according to the reserve prediction accuracy and the mixed matrix index.
In a target investment prediction model for new energy automobile investment prediction, the XGB OST model can be determined as an optimal investment prediction model under the conditions of large quantity and/or sparseness of features to be trained according to investment prediction accuracy and a mixed Matrix index (fusion Matrix).
And determining the optimal investment reservation prediction model in the target investment reservation prediction model as the XGBOOST model.
In the process of determining an optimal investment reservation prediction model suitable for investment reservation prediction in at least two investment reservation prediction models as a target investment reservation prediction model, as shown in fig. 2d, the method includes the following steps:
step 23031: and iterating the features to be trained in the training sample.
Step 23032: and when the accuracy of the target refund prediction is higher than a first preset value, obtaining a refund prediction characteristic.
And determining the investment retention prediction characteristic by further judging the investment retention prediction accuracy obtained by the target investment retention prediction model, and further arranging information delivery or advertisement delivery of the public and private domain platform by the determined investment retention prediction characteristic to improve the accuracy of investment retention prediction.
According to the embodiment of the invention, all user prediction characteristics contained in public and private domain data can be integrated by acquiring the user prediction characteristics comprising the user characteristics in the public domain data and the user characteristics in the private domain data; the user prediction features can determine the features to be trained related to the reserve resource prediction through a preset screening rule; training through a training sample containing the characteristics to be trained to obtain a target investment prediction model; training at least two funding prediction models based on a training sample containing the characteristics to be trained; obtaining the withholding prediction accuracy and the mixed matrix index output by at least two withholding prediction models; according to the reserve prediction accuracy and the mixed matrix index, determining an optimal reserve prediction model suitable for reserve prediction in at least two reserve prediction models as a target reserve prediction model, and iterating the characteristics to be trained in the training sample; when the accuracy of the target refund prediction is higher than a first preset value, obtaining a refund prediction characteristic; all user prediction characteristics contained in the public and private domain data can be integrated; the reserve can be predicted more accurately.
FIG. 3 is a flow chart illustrating another embodiment of the user funding prediction training method of the present invention, which is performed by a user funding prediction device. As shown in fig. 3, the method comprises the steps of:
step 310: acquiring user prediction characteristics; wherein the user predicted characteristics include user characteristics in public domain data and user characteristics in private domain data.
The public domain data is data of a public domain flow platform obtained by directly entering a mature platform for automobile enterprises (vehicle enterprises for short) to realize flow conversion. The private domain data is user characteristics obtained by a user in an automobile brand private domain, and also can be user characteristics obtained in a new energy automobile brand private domain, wherein the private domain can comprise channels such as a brand official website, a brand APP (Application, mobile phone Application), a brand applet and the like.
The user characteristics include user attributes and user behavior.
The user attributes may include user basic information, vehicle purchasing intention and/or congestion measurement system values. The basic information of the user can comprise gender, age bracket, ordinary station, family resident population, income grade, whether a vehicle exists or not and the like; the car buying intent may include vehicle type preferences, energy preferences and/or car buying budget; the user metric system values may include an activity force value, an influence force value, a consumption force value, and/or a total user metric value.
The user behavior of the public domain data includes advertisement exposure and advertisement click, and may include a media platform, an advertisement type, a content type, and an exposure number of N days or YTD.
The user behaviors of the private domain data comprise behaviors on an official website, an applet and/or an application program, and the official website behaviors can comprise browsing days in N days and video playing number and frequency in N days; the small program behavior comprises whether a mobile phone number is authorized or not, the login days and times in N days, and the video playing numbers and times in N days; the application behavior comprises registration, login days and login times in N days, video playing numbers and video playing times in N days, and browsing post numbers and comment numbers in N days.
The user behavior is not limited to the specific behaviors listed above, but also includes other behaviors that may trigger user behavior, possibly resulting in user funding.
Step 320: and determining the characteristics to be trained related to the investment prediction by the user prediction characteristics through a preset screening rule.
The preset screening rule may be Chi-Squared Test (Chi-Squared Test), or may be a Test of other statistical classification inference methods.
And analyzing the characteristic correlation of the user prediction characteristic through chi-square test to determine the characteristic to be trained.
Among them, chi-Squared Test (Chi-Squared Test) subjects the distribution of statistics to a hypothesis Test that approximates a Chi-Squared distribution when a null hypothesis holds. The chi-square test may refer to the pearson chi-square test, wherein the chi-square test is the degree of deviation between the actual observed value and the theoretical inferred value of the statistical sample.
In step 320, the method comprises:
obtaining statistics of the user predicted features and degrees of freedom of the statistics.
Wherein the test is carried out by the Chi-Square test, in particular the Pearson Chi-Square test. The statistical value is obtained by dividing the difference between the observed value and the theoretical value of the user prediction characteristic by the theoretical value after squaring the difference, and then summing the values. The degree of freedom can be obtained by statistical values.
And determining the characteristic to be trained according to the statistical value, the degree of freedom and the linear correlation of the Pearson correlation coefficient.
The feature to be trained can be determined according to a chi-square distribution critical value and a statistic value of the degree of freedom and the square of the Pearson correlation coefficient r.
Step 330: and training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
The method comprises the following steps of training through a training sample containing the characteristics to be trained to obtain a target resource reservation prediction model, wherein the method comprises the following steps: and training at least two investment prediction models based on the training samples containing the features to be trained. And acquiring the remaining prediction accuracy and the mixed matrix index output by at least two remaining prediction models. And determining an optimal remaining resource prediction model suitable for remaining resource prediction in at least two remaining resource prediction models as a target remaining resource prediction model according to the remaining resource prediction accuracy and the mixed matrix index.
Determining an optimal resource reservation prediction model suitable for resource reservation prediction in at least two resource reservation prediction models as a target resource reservation prediction model, and iterating the features to be trained in the training samples; and when the accuracy of the target refund prediction is higher than a first preset value, obtaining a refund prediction characteristic.
The at least two persistent prediction models may include at least two models of a Logistic Regression model (Logistic Regression), a Random Forest Regression model (Random Forest), a Gradient Boosting Tree model (GBDT), and/or a Support Vector Machine (SVM), and may further include other integrated learning models besides the above listed models.
In a target investment prediction model for new energy automobile investment prediction, the XGB OST model can be determined as an optimal investment prediction model under the conditions of large quantity and/or sparseness of features to be trained according to investment prediction accuracy and a mixed Matrix index (fusion Matrix).
And determining the optimal investment reservation prediction model in the target investment reservation prediction model as the XGBOOST model.
And determining the investment retention prediction characteristic by further judging the investment retention prediction accuracy obtained by the target investment retention prediction model, and further arranging information delivery or advertisement delivery of the public and private domain platform by the determined investment retention prediction characteristic to improve the accuracy of investment retention prediction.
Step 340: the method comprises the steps of popularizing for a first group of users and a second group of users, and determining the remaining capital conversion rate of the first group of users and the remaining capital conversion rate of the second group of users; the first group of users are users of a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users of the first number, randomly selected from the promoted users.
The number of the users of the second group is equal to the number of the users of the first group, and the number of the users of the first group and the number of the users of the second group are both a first number. The first group of users and the second group of users can be compared according to the advertisement exposure, the click number and the remaining data of the user groups in the public and private domains.
The refuge conversion rate is obtained by dividing the number of the refuge users by the number of the users in the group, so that the refuge conversion rate of the first group of users is the number of the refuge users of the first group of users divided by the number of the users of the first group of users; the refuge conversion rate of the second group of users is the number of refuge users of the second group of users divided by the number of users of the second group of users.
Step 350: and if the capital-saving conversion rate of the first group of users is higher than the capital-saving conversion rate of the second group of users, or the capital-saving conversion rate of the first group of users is higher than a third preset value, taking the target capital-saving prediction model as a final application capital-saving prediction model.
And comparing and verifying the first group of users and the second group of users to further determine the withholding prediction characteristic as the current optimal characteristic, thereby improving the withholding prediction of the vehicle enterprises through the public and private domain data.
Step 360: and if the capital-remaining conversion rate of the first group of users is lower than or equal to the capital-remaining conversion rate of the second group of users, or the capital-remaining conversion rate of the first group of users is lower than or equal to a third preset value, optimizing the target capital-remaining prediction model through the characteristics to be trained.
In some embodiments, when the funding conversion rate of the first group of users is lower than or equal to the funding conversion rate of the second group of users, model parameters of an optimal funding prediction model in the target funding prediction model are adjusted, and new funding prediction features are determined according to the optimal funding prediction model with the adjusted parameters.
In some embodiments, when the funding conversion rate of the first group of users is less than or equal to a third preset value, model parameters of an optimal funding prediction model in the target funding prediction model are adjusted, and new funding prediction characteristics are determined according to the optimal funding prediction model with the adjusted parameters.
And performing a comparison experiment according to the new reserve forecasting characteristics until the current optimal characteristics are obtained, and improving the reserve forecasting of the vehicle enterprises through the public and private domain data.
The behavior of the user in the private domain and the accuracy of data on the funding prediction are higher than those in the public domain, and the prediction on the desirability of the funding is more important.
By the method, the capital remaining probability of each user can be accurately calculated, and the accuracy of targeted advertisement delivery is improved.
The embodiment of the invention obtains the user prediction characteristics comprising the user characteristics in the public domain data and the user characteristics in the private domain data, and can integrate all the user prediction characteristics contained in the public and private domain data; the user prediction features can determine the features to be trained related to the investment prediction through a preset screening rule; training through a training sample containing the characteristics to be trained to obtain a target investment prediction model; the method comprises the steps of popularizing for a first group of users and a second group of users, and determining the remaining capital conversion rate of the first group of users and the remaining capital conversion rate of the second group of users; the first group of users are users with a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users with the first number, randomly selected from the promoted users; all user prediction characteristics contained in the public and private domain data can be integrated; the reserve can be predicted more accurately.
Fig. 4 is a schematic structural diagram of an embodiment of the user investment prediction apparatus of the present invention. As shown in fig. 4, the user investment prediction apparatus 400 includes: an acquisition module 410, a verification module 420, and a model module 430.
An obtaining module 410, configured to obtain a user prediction feature; wherein the user predicted characteristics comprise user characteristics in public domain data and user characteristics in private domain data; the user characteristics include user attributes and user behavior.
The user attributes may include user basic information, car buying intent, and/or congestion metric system values, among others. The basic information of the user can comprise gender, age bracket, frequent residence, family resident population, income grade, whether a car exists or not and the like; the car buying intent may include vehicle type preferences, energy preferences and/or car buying budget; the user metric system values may include an activity force value, an influence force value, a consumption force value, and/or a total user metric value.
The user behavior of the public domain data includes advertisement exposure and advertisement click, and may include a media platform, an advertisement type, a content type, and an exposure number of N days or YTD.
The user behaviors of the private domain data comprise behaviors on an official website, an applet and/or an application program, and the official website behaviors can comprise browsing days in N days and video playing numbers and times in N days; the small program behavior comprises whether a mobile phone number is authorized, the number of login days and the number of login times in N days, and the number of video playing times in N days; the application behavior comprises registration, login days and login times in N days, video playing numbers and video playing times in N days, and browsing post numbers and comment numbers in N days.
The user behavior is not limited to the specific behaviors listed above, but also includes other behaviors that may trigger user behavior, possibly resulting in user funding.
And the checking module 420 is used for determining the characteristics to be trained related to the reserve prediction by the user prediction characteristics through a preset screening rule.
And the model module 430 is configured to train through the training sample including the feature to be trained to obtain a target investment prediction model.
In an alternative, the model module is further configured to:
training at least two investment prediction models based on the training sample containing the characteristics to be trained;
obtaining the withholding prediction accuracy and the mixed matrix index output by at least two withholding prediction models;
and determining an optimal reserve prediction model suitable for reserve prediction in at least two reserve prediction models as a target reserve prediction model according to the reserve prediction accuracy and the mixed matrix index.
In an optional manner, the model module is further configured to:
iterating the features to be trained in the training samples;
and when the accuracy of the reserve prediction output by the target reserve prediction model is higher than a preset value, taking the feature to be trained in the training sample as a reserve prediction feature.
In an optional manner, the preset filtering rule is chi-square test, and the test module is further configured to:
obtaining a statistic of the user prediction characteristic and a degree of freedom of the statistic;
and determining the characteristic to be trained related to the vestigial prediction according to the statistic value, the degree of freedom and the linear correlation of the Pearson correlation coefficient.
The refund prediction device also comprises a verification module, a data processing module and a data processing module, wherein the verification module is used for popularizing a first group of users and a second group of users and determining the refund conversion rate of the first group of users and the refund conversion rate of the second group of users; the first group of users are users with a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users with the first number, randomly selected from the promoted users.
In an optional manner, after determining the funding conversion rate of the first group of users and the funding conversion rate of the second group of users for the first group of users and the second group of users, the verification module is further configured to:
and if the capital-saving conversion rate of the first group of users is higher than the capital-saving conversion rate of the second group of users, or the capital-saving conversion rate of the first group of users is higher than a third preset value, taking the target capital-saving prediction model as a final application capital-saving prediction model.
In an optional manner, after determining the funding conversion rate of the first group of users and the funding conversion rate of the second group of users for the first group of users and the second group of users, the verification module is further configured to:
and if the capital-remaining conversion rate of the first group of users is lower than or equal to the capital-remaining conversion rate of the second group of users, or the capital-remaining conversion rate of the first group of users is lower than or equal to a third preset value, optimizing the target capital-remaining prediction model through the characteristics to be trained.
The embodiment of the invention is used for acquiring the user prediction characteristics through the acquisition module; wherein the user predicted characteristics comprise user characteristics in public domain data and user characteristics in private domain data; the checking module is used for determining the characteristics to be trained related to the reserve resources prediction by the user prediction characteristics through a preset screening rule; and the model module is used for training through the training sample containing the characteristics to be trained to obtain a target investment prediction model. All user prediction characteristics contained in the public and private domain data can be integrated; the reserve can be predicted more accurately.
An embodiment of the present invention provides a computer-readable storage medium, where the storage medium stores at least one executable instruction, and when the executable instruction runs on a user investment prediction apparatus, the user investment prediction apparatus executes a user investment prediction training method in any method embodiment described above.
The executable instructions may be specifically configured to cause the user funding prediction device to:
acquiring user prediction characteristics; wherein the user prediction features comprise user features in public domain data and user features in private domain data;
the user prediction features determine the features to be trained related to the investment prediction through a preset screening rule;
and training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
In an optional manner, the training by using the training sample including the feature to be trained to obtain the target funding prediction model further includes:
training at least two investment prediction models based on the training sample containing the characteristics to be trained;
obtaining the capital-reserving prediction accuracy rate and the mixed matrix index output by at least two capital-reserving prediction models;
and determining an optimal remaining resource prediction model suitable for remaining resource prediction in at least two remaining resource prediction models as a target remaining resource prediction model according to the remaining resource prediction accuracy and the mixed matrix index.
In an optional manner, the training by using the training sample including the feature to be trained to obtain the target funding prediction model further includes:
iterating the features to be trained in the training samples;
and when the qualification forecasting accuracy output by the target qualification forecasting model is higher than a preset value, taking the feature to be trained in the training sample as a qualification forecasting feature.
In an optional manner, after the training is performed through the training sample including the feature to be trained to obtain the target funding prediction model, the method further includes:
popularizing aiming at a first group of users and a second group of users, and determining the refunding conversion rate of the first group of users and the refunding conversion rate of the second group of users; the first group of users are users of a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users of the first number, randomly selected from the promoted users.
In an optional manner, after determining the funding conversion rate of the first group of users and the funding conversion rate of the second group of users for the first group of users and the second group of users, the method further includes:
and if the capital-saving conversion rate of the first group of users is higher than the capital-saving conversion rate of the second group of users, or the capital-saving conversion rate of the first group of users is higher than a third preset value, taking the target capital-saving prediction model as a final application capital-saving prediction model.
In an optional manner, after determining the funding conversion rate of the first group of users and the funding conversion rate of the second group of users for the first group of users and the second group of users, the method further includes:
and if the capital-remaining conversion rate of the first group of users is lower than or equal to the capital-remaining conversion rate of the second group of users, or the capital-remaining conversion rate of the first group of users is lower than or equal to a third preset value, optimizing the target capital-remaining prediction model through the characteristics to be trained.
In an optional manner, the preset filtering rule is chi-square test, and the user predicted feature determines a feature to be trained related to the funding prediction according to the preset filtering rule, further including:
obtaining a statistic of the user prediction characteristic and a degree of freedom of the statistic;
and determining the characteristic to be trained related to the vestigial prediction according to the statistic value, the degree of freedom and the linear correlation of the Pearson correlation coefficient.
According to the embodiment of the invention, all the user prediction characteristics contained in the public and private domain data can be integrated by acquiring the user prediction characteristics comprising the user characteristics in the public domain data and the user characteristics in the private domain data; through chi-square test, the characteristics to be trained can be determined; the training sample containing the characteristics to be trained passes through the target fund reserving prediction model, so that the fund reserving prediction accuracy can be obtained; through the investment prediction accuracy, investment prediction characteristics can be determined; the reserve can be predicted more accurately.
The algorithms or displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus. In addition, embodiments of the present invention are not directed to any particular programming language.
In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. Similarly, in the above description of exemplary embodiments of the invention, various features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the invention and aiding in the understanding of one or more of the various inventive aspects. Where the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components in the embodiments may be combined into one module or unit or component, and furthermore, may be divided into a plurality of sub-modules or sub-units or sub-components. Except that at least some of such features and/or processes or elements are mutually exclusive.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means can be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names. The steps in the above embodiments should not be construed as limiting the order of execution unless specified otherwise.
Claims (10)
1. A user funding prediction training method, the method comprising:
acquiring user prediction characteristics; wherein the user prediction features comprise user features in public domain data and user features in private domain data;
the user prediction features determine features to be trained related to the reserve resource prediction through a preset screening rule;
and training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
2. The user funding prediction training method according to claim 1, wherein the training is performed by using a training sample containing the feature to be trained to obtain a target funding prediction model, and further comprising:
training at least two funding prediction models based on a training sample containing the features to be trained;
obtaining the capital-reserving prediction accuracy rate and the mixed matrix index output by at least two capital-reserving prediction models;
and determining an optimal remaining resource prediction model suitable for remaining resource prediction in at least two remaining resource prediction models as a target remaining resource prediction model according to the remaining resource prediction accuracy and the mixed matrix index.
3. The user funding prediction training method according to claim 2, wherein the determining an optimal funding prediction model suitable for funding prediction from among the at least two funding prediction models as a target funding prediction model further comprises:
iterating the features to be trained in the training samples;
and when the accuracy of the reserve prediction output by the target reserve prediction model is higher than a first preset value, taking the feature to be trained in the training sample as a reserve prediction feature.
4. The user funding prediction training method according to claim 1, wherein after training through the training sample containing the feature to be trained to obtain a target funding prediction model, the method further comprises:
the method comprises the steps of popularizing for a first group of users and a second group of users, and determining the remaining capital conversion rate of the first group of users and the remaining capital conversion rate of the second group of users; the first group of users are users with a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users with the first number, randomly selected from the promoted users.
5. The user funding prediction training method of claim 4, wherein after determining the funding conversion rate of the first group of users and the funding conversion rate of the second group of users for the first group of users and the second group of users, the method further comprises:
and if the capital-saving conversion rate of the first group of users is higher than the capital-saving conversion rate of the second group of users, or the capital-saving conversion rate of the first group of users is higher than a third preset value, taking the target capital-saving prediction model as a final application capital-saving prediction model.
6. The user funding prediction training method of claim 4, wherein after determining the funding conversion rate for the first group of users and the funding conversion rate for the second group of users for the first group of users and the second group of users, the method further comprises:
and if the capital-remaining conversion rate of the first group of users is lower than or equal to the capital-remaining conversion rate of the second group of users, or the capital-remaining conversion rate of the first group of users is lower than or equal to a third preset value, optimizing the target capital-remaining prediction model through the characteristics to be trained.
7. The user funding prediction training method according to claim 1, wherein the preset filtering rule is chi-square test, and the user predicted feature determines the feature to be trained related to the funding prediction through the preset filtering rule, further comprising:
obtaining a statistic of the user prediction characteristic and a degree of freedom of the statistic;
and determining the characteristic to be trained related to the vestigial prediction according to the statistic value, the degree of freedom and the linear correlation of the Pearson correlation coefficient.
8. A user funding prediction apparatus, comprising:
the acquisition module acquires user prediction characteristics; wherein the user prediction features comprise user features in public domain data and user features in private domain data;
the checking module is used for determining the characteristics to be trained related to the reserve resources prediction by the user prediction characteristics through a preset screening rule;
and the model module is used for training through the training sample containing the characteristics to be trained to obtain a target investment prediction model.
9. The user funding prediction device according to claim 8, wherein the user funding prediction device further comprises:
the verification module is used for popularizing aiming at a first group of users and a second group of users and determining the refund conversion rate of the first group of users and the refund conversion rate of the second group of users; the first group of users are users of a first number, predicted by the target fund reserving prediction model, of promoted users, and the fund reserving prediction value of the promoted users is higher than a second preset value, and the second group of users are users of the first number, randomly selected from the promoted users.
10. A computer-readable storage medium having stored thereon at least one executable instruction that, when executed on a user funding prediction device, causes the user funding prediction device to perform operations of a user funding prediction training method as recited in any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210959399.0A CN115330447A (en) | 2022-08-10 | 2022-08-10 | User investment prediction training method, device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210959399.0A CN115330447A (en) | 2022-08-10 | 2022-08-10 | User investment prediction training method, device and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115330447A true CN115330447A (en) | 2022-11-11 |
Family
ID=83920900
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210959399.0A Pending CN115330447A (en) | 2022-08-10 | 2022-08-10 | User investment prediction training method, device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115330447A (en) |
-
2022
- 2022-08-10 CN CN202210959399.0A patent/CN115330447A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108121795B (en) | User behavior prediction method and device | |
| CN110826006B (en) | Abnormal collection behavior identification method and device based on privacy data protection | |
| US8346710B2 (en) | Evaluating statistical significance of test statistics using placebo actions | |
| CN107862339B (en) | Method and apparatus for outputting information | |
| CN108399482B (en) | Contract evaluation method and device and electronic equipment | |
| CN111383101A (en) | Post-loan risk monitoring method, device, equipment and computer-readable storage medium | |
| CN108876188B (en) | Inter-connected service provider risk assessment method and device | |
| US11636354B1 (en) | System and method for managing social-based questions and answers | |
| CN112529575B (en) | Risk early warning method, equipment, storage medium and device | |
| CN113407854A (en) | Application recommendation method, device and equipment and computer readable storage medium | |
| CN113393299A (en) | Recommendation model training method and device, electronic equipment and storage medium | |
| CN111061948B (en) | User tag recommendation method and device, computer equipment and storage medium | |
| CN112184143B (en) | Model training method, device and equipment in compliance audit rule | |
| CN113919432A (en) | Classification model construction method, data classification method and device | |
| CN110334936B (en) | Method, device and equipment for constructing credit qualification scoring model | |
| CN112836128A (en) | Information recommendation method, device, equipment and storage medium | |
| CN111582647A (en) | User data processing method and device and electronic equipment | |
| CN115330447A (en) | User investment prediction training method, device and computer readable storage medium | |
| CN111160797A (en) | Wind control model construction method and device, storage medium and terminal | |
| Hoover et al. | Understanding crash risk using a multi-level random parameter binary logit model: application to naturalistic driving study data | |
| KR102423514B1 (en) | Method of operating used car trading platform that supports direct transactions between sellers and buyers, in which experts directly check accident history, complete on-site inspections, inform inspection results, and provide follow-up management services with self-assurance | |
| CN113139182B (en) | Data intrusion detection method for online e-commerce platform | |
| CN114493250A (en) | Abnormal behavior detection method, computing device and readable storage medium | |
| CN112016964A (en) | Second-hand vehicle dynamic pricing method and device, electronic equipment and storage medium | |
| US11822562B2 (en) | Unstructured text processing for geographical location determination system |
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 |