CN114238763A - Information pushing method and device based on financial APP and computer equipment - Google Patents

Information pushing method and device based on financial APP and computer equipment Download PDF

Info

Publication number
CN114238763A
CN114238763A CN202111557181.4A CN202111557181A CN114238763A CN 114238763 A CN114238763 A CN 114238763A CN 202111557181 A CN202111557181 A CN 202111557181A CN 114238763 A CN114238763 A CN 114238763A
Authority
CN
China
Prior art keywords
information
user
model
training
pushed
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
Application number
CN202111557181.4A
Other languages
Chinese (zh)
Inventor
李恒逸
王栋
余星梅
向阳
张超
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Industrial and Commercial Bank of China Ltd ICBC
Original Assignee
Industrial and Commercial Bank of China Ltd ICBC
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Industrial and Commercial Bank of China Ltd ICBC filed Critical Industrial and Commercial Bank of China Ltd ICBC
Priority to CN202111557181.4A priority Critical patent/CN114238763A/en
Publication of CN114238763A publication Critical patent/CN114238763A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • 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
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
    • 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
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/06—Asset management; Financial planning or analysis

Abstract

The application relates to an information pushing method, an information pushing device, computer equipment, a storage medium and a computer program product based on financial APP. The method comprises the following steps: acquiring training samples, wherein each training sample comprises a plurality of training characteristics; performing feature processing on the training features to obtain processed training features; acquiring a user calling model based on the processed training characteristics and the logistic regression model; the method comprises the steps of obtaining information of a user to be pushed, obtaining a user with a high probability of calling based on the information of the user to be pushed and a user calling model, and pushing the information to the user with the high probability of calling. By adopting the method, the call-to-alive efficiency of information push of the user can be improved.

Description

Information pushing method and device based on financial APP and computer equipment
Technical Field
The present application relates to the field of internet finance technologies, and in particular, to an information pushing method and apparatus based on a financial APP, a computer device, a storage medium, and a computer program product.
Background
The user operation of the financial APP is mainly carried out through information push at present, and operators configure information push information at the background and display the information push information in a message notification bar of a mobile phone of the user and a message communication tab of the financial APP after the information push information is completed. In the aspect of pushing crowds, marketing personnel generally push the whole amount of users at present, and the pertinence is not strong; in the aspect of pushing parameters, the pushing time and the pushing content are set according to the experience of related marketing personnel in the early stage, and the optimal pushing effect cannot be achieved.
Financial APP current propelling movement is mostly whole propelling movement, can't carry out the pertinence marketing according to user's characteristics, leads to a lot of marketing resources to have gone down to the user group that can not be called live, leads to calling the result after the propelling movement unsatisfactory, has caused the waste of marketing resources, and holistic propelling movement efficiency is on the low side, has caused unnecessary disturbance to non-target user.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a financial APP-based information pushing method, apparatus, computer device, computer-readable storage medium, and computer program product, which can improve call-after-push efficiency.
In a first aspect, the application provides an information pushing method based on a financial APP. The method comprises the following steps:
acquiring training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In one embodiment, the obtaining a user arousal model based on the processed training features and the logistic regression model includes:
acquiring a first arousing model based on the processed training features and the logistic regression model;
performing principal component analysis processing on the processed training features to obtain the training features after principal component analysis;
acquiring a second revival model based on the training characteristics and the logistic regression model after the principal component analysis;
and acquiring a model with high probability of calling a user from the first calling model and the second calling model as a user calling model.
In one embodiment, the obtaining a second arousal model based on the training features and the logistic regression model after the principal component analysis comprises:
and training the logistic regression model based on the training characteristics after the principal component analysis, and obtaining the trained logistic regression model as a second arousing model.
In one embodiment, the obtaining a model with a high probability of evoking a user from the first evoking model and the second evoking model as a user evoking model comprises:
selecting AUC and ROC as evaluation indexes of the trained logistic regression model;
and respectively evaluating the first and second reviving models based on the evaluation indexes, and acquiring a model with high effect evaluation in the first and second reviving models as the user reviving model.
In one embodiment, the obtaining of the user with the high arousal probability based on the information of the user to be pushed and the user arousal model includes:
inputting the information of the users to be pushed to the user call-alive model, and acquiring the call-alive probability of each user to be pushed;
and screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
In one embodiment, said pushing information to said high arousal probability user comprises:
selecting a functional scene-based file title as an information title to be pushed, and simultaneously selecting high-welfare activities as information contents to be pushed;
and pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
In a second aspect, the application further provides an information pushing device based on the financial APP. The device comprises:
the device comprises a sample acquisition module, a data acquisition module and a data processing module, wherein the sample acquisition module is used for acquiring training samples, and each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
the characteristic processing module is used for carrying out characteristic processing on the training characteristics to obtain the processed training characteristics; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
the model acquisition module is used for acquiring a user calling model based on the processed training characteristics and the logistic regression model;
the information pushing module is used for acquiring user information to be pushed, acquiring a user with a high call probability based on the user information to be pushed and the user call model, and pushing information to the user with the high call probability; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the following steps when executing the computer program:
acquiring training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
acquiring training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program which when executed by a processor performs the steps of:
acquiring training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
According to the information pushing method and device based on the financial APP, the training samples are obtained, each training sample comprises a plurality of training features, the training features are subjected to feature processing, the processed training features are obtained, the user call model is obtained based on the processed training features and the logical regression model, the user information to be pushed is obtained, the users with high call probability are obtained based on the user information to be pushed and the user call model, the information is pushed to the users with high call probability, the user call model is generated through training, the users with high call probability are screened out to carry out information pushing based on the user call model, and the pushing call success rate of the users is improved.
Drawings
FIG. 1 is a diagram of an application environment of an information pushing method based on a financial APP in one embodiment;
FIG. 2 is a flow chart illustrating an information pushing method based on a financial APP in one embodiment;
FIG. 3 is a flowchart illustrating the user step of obtaining a high probability of arousal in one embodiment;
FIG. 4 is a block diagram of an information pushing apparatus based on a financial APP 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 information pushing method based on the financial APP provided by the embodiment of the application can be applied to the application environment shown in FIG. 1. Wherein the terminal 102 communicates with the server 104 via a network. The data storage system may store data that the server 104 needs to process. The data storage system may be integrated on the server 104, or may be located on the cloud or other network server.
The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, internet of things devices and portable wearable devices, and the internet of things devices may be smart speakers, smart televisions, smart air conditioners, smart car-mounted devices, and the like. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, and the like. The server 104 may be implemented as a stand-alone server or as a server cluster comprised of multiple servers.
In one embodiment, as shown in fig. 2, there is provided an information pushing method based on financial APP, which is described by taking the method as an example for being applied to the server in fig. 1, and includes the following steps:
step 202, obtaining training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information.
The user registration information is the most basic characteristic for marking a client, and according to the statistics of industry data, the response degree of people with different ages and sexes to the internet products is greatly different. The method mainly comprises the following steps: age, gender, education, occupation, marital status, nationality, unit nature, annual income, date of registration, etc. The user asset information mainly comprises: personal gross assets, deposit balances, national bond balances, fund balances, bond balances, various loan balances, and the like. The service provisioning information mainly includes: the mobile banking system comprises a mobile banking opening identification, a financing opening identification, a wealth card opening identification, a public deposit opening identification, a (quasi-) credit card opening identification, an open fund opening identification, a mobile account number and the like. The key function usage information mainly includes: the latest login time, the click condition of the accounting book, the click condition of friend transfer, the click condition of deposited money, the click condition of point exchange, and the like. The region click information mainly comprises: the click rate of each atomic function in the home high-frequency area and the click rate of each atomic function in the home self-defined function area, the completion of accounting, the entry of a home message center, the completion of searching, the click of a home tab, the click of a scene number tab, the click of my tab and the like. The user transaction information mainly comprises: comprehensive account transaction conditions, and the like. The internet product use information mainly comprises: the number of internet bank logins, and the like.
Specifically, training samples are obtained, wherein each training sample comprises a plurality of training characteristics; the training samples are selected from all users pushed by the financial APP at a certain time, and a preset number of users are selected according to a random sampling rule to serve as the training samples of the model. Each training sample comprises at least one training characteristic, and the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information.
Step 204, performing feature processing on the training features to obtain processed training features; the process of feature processing comprises at least one of missing value processing, variable derivation, data normalization, and categorical variable processing.
Specifically, performing feature processing on the training features to obtain processed training features; the process of feature processing includes at least one of missing value processing, variable derivation, data normalization, and categorical variable processing. The missing value processing means that for variables with missing quantity exceeding 80%, the effective content is less, so that the client group cannot be effectively distinguished and eliminated. The variable derivation is to derive a new index according to the existing index and the actual service scene; for example, for the days from the last login time to the present, the days from the last login to the present are calculated according to the last login date and used as the response index of the sleeping condition of the user; for the nationality of the user, as most samples are Chinese clients and foreign clients are sparse, the dictionary values of the nationality code fields are divided again, so that the characteristics of the region to which the user belongs are represented better; calculating the days of creation till now according to the user account creation time for the days of creation till now, wherein the days of creation till now is used as an identification index of the user use time; for the login times of the financial APP in the T-X month, as the financial APP belongs to the Internet product, the login condition of the financial APP can reflect the use condition of the Internet product accepted by the user, and therefore, the login log data of the financial APP in the T-X month is derived and used for identifying the active condition of the financial APP of the user; and for the transaction times of the T-X month, deriving a transaction time index through a transaction log table, and identifying the transaction activity degree of the user.
In the data standardization processing, in the real modeling, some numerical characteristics have dimension difference, so different characteristic variables have the same scale through the standardization processing, the difference between the characteristics is eliminated, and the model can conveniently learn the characteristic weight. For numerical variables, the conversion methods for normalization include centering conversion (making the mean of the variables 0), normalization conversion (making the mean of the variables 0 and the standard deviation 1), and range conversion (making the minimum of the variables 0 and the maximum of the variables 1). Most of numerical variables in the original data are extremely poor (such as deposit balance and the like), so that the data are concentrated by using extremely poor conversion and the model training is not facilitated; normalization transforms work better than centralized transforms, so the numerical variable normalization herein uses a method of normalizing transforms. The principle is as follows:
the number of the numerical sample is n, each sample has m observation indexes, and each observation value is marked as Xij(i 1, 2.. times., n; j 1, 2.. times., m), average value
Figure BDA0003419259940000071
Standard deviation of
Figure BDA0003419259940000072
And (3) standardization treatment process:
Figure BDA0003419259940000073
in this embodiment, a STANDATD process provided by the SAS is used to perform numerical data normalization, and the processed data conforms to a standard normal distribution.
The classification variable processing is a variable which has only a label value and no numerical value, and has no size relation between classifications. Firstly, integer coding is carried out on classified variables, and classification is represented by numbers such as 0,1, 2. one-hot coding can remove integer coding, a binary variable is created for each integer value, the binary variables are mutually exclusive and are generally called dummy variables, and the classified variables after one-hot processing can be learned by a machine learning model.
And step 206, acquiring a user arousing model based on the processed training characteristics and the logistic regression model.
Specifically, the research objective of the application is to predict the acceptance degree of financial APP users to push, so as to optimize crowd circle selection, select the user group which is most likely to be called for life for marketing push, and essentially is a typical classification problem, so that logistic regression and decision tree models are selected for analysis. And training the logistic regression model based on the processed training characteristics to obtain a trained user arousing model.
Step 208, obtaining user information to be pushed, obtaining users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
Specifically, after the user call model is obtained, the information of the user to be pushed is obtained, and the information type in the information of the user to be pushed is the same as the information type in the training feature. Inputting the information of the user to be pushed into the user call model, acquiring the call probability of the user to be pushed, screening the call probability of the user through a preset threshold value, screening the user with high call probability, and pushing the information to the screened user with high call probability.
According to the information pushing method based on the financial APP, training samples are obtained, each training sample comprises a plurality of training features, the training features are subjected to feature processing, the processed training features are obtained, a user call model is obtained based on the processed training features and a logistic regression model, user information to be pushed is obtained, users with high call probability are obtained based on the user information to be pushed and the user call model, information is pushed to the users with high call probability, the user call model is generated through training, the users with high call probability are screened out based on the user call model to carry out information pushing, and the push call success rate of the users is improved.
In one embodiment, the obtaining a user arousal model based on the processed training features and the logistic regression model includes:
acquiring a first arousing model based on the processed training features and the logistic regression model;
performing principal component analysis processing on the processed training features to obtain the training features after principal component analysis;
acquiring a second revival model based on the training characteristics and the logistic regression model after the principal component analysis;
and acquiring a model with high probability of calling a user from the first calling model and the second calling model as a user calling model.
In particular, multivariate large data sets undoubtedly provide abundant information for research and application, but also increase the workload of data acquisition to some extent. More importantly, in many cases, there may be correlations between many variables, thereby increasing the complexity of problem analysis. If each index is analyzed separately, the analysis is often isolated and cannot fully utilize the information in the data, so blindly reducing the index will lose much useful information, thereby leading to erroneous conclusions. Dimension reduction is a method for preprocessing high-dimensional feature data. Dimension reduction is to retain data with high dimension with some most important features, and remove noise and unimportant features, thereby achieving the purpose of increasing data processing speed. In actual production and application, dimension reduction can save a great deal of time and cost for people within a certain information loss range. Dimension reduction also becomes a very widely applied data preprocessing method.
Principal Component Analysis (PCA) is one of the most widely used data dimension reduction algorithms. The main idea of PCA is to map n-dimensional features onto k-dimensions, which are completely new orthogonal features, also called principal components, and k-dimensional features reconstructed on the basis of the original n-dimensional features. The task of PCA is to sequentially find a set of mutually orthogonal axes from the original space, the selection of new axes being strongly dependent on the data itself. The first new coordinate axis is selected to be the direction with the largest square difference in the original data, the second new coordinate axis is selected to be the plane which is orthogonal to the first coordinate axis and enables the square difference to be the largest, the third axis is the plane which is orthogonal to the 1 st axis and the 2 nd axis and has the largest square difference, and the like are carried out in sequence.
And carrying out PCA by using PRINCOMP process steps provided by SAS in the process of obtaining the second arousing model based on the training characteristics and the logistic regression model after principal component analysis, wherein the main process is as follows:
1. a covariance matrix of the samples is calculated.
2. And solving the eigenvalue of the covariance matrix and the corresponding eigenvector.
3. And arranging the eigenvectors into a matrix from top to bottom according to the size of the corresponding eigenvalue, and taking the first k rows to form a matrix p.
4. And Y is PX which is the data from dimensionality reduction to dimensionality K.
The PCA has the following advantages after use: (1) the information amount only needs to be measured by the variance and is not influenced by factors except the data set. (2) The main components are orthogonal, so that the factors influencing each other among the original data components can be eliminated. (3) And selecting the combination with most information in the characteristics to reduce the variable dimensionality.
After the SAS is used for PCA operation, the variable dimensionality can be reduced, and the subsequent model training is simplified. Based on the initial screening of clustering results, a machine learning method is used for dynamically selecting the logic modeling effect under the data which is subjected to PCA and does not use PCA dimension reduction processing, an optimal feature set is output, whether users are called for life or not is predicted, and finally a calling model is obtained and used for calculating the probability of calling for life of each user, so that the subsequent pushed crowd selection is facilitated, and accurate marketing is realized.
In this embodiment, a first revival model is obtained based on the processed training features and the logistic regression model, principal component analysis is performed on the processed training features, the training features after the principal component analysis are obtained, a second revival model is obtained based on the training features and the logistic regression model after the principal component analysis, and finally a model with a high probability of reviving a user in the first revival model and the second revival model is obtained as a user revival model.
In one embodiment, the obtaining a second arousal model based on the principal component analyzed training features and the logistic regression model includes:
and training the logistic regression model based on the training characteristics after the principal component analysis, and obtaining the trained logistic regression model as a second arousing model.
Specifically, the logistic regression model is trained based on the training features after principal component analysis, and the trained logistic regression model is obtained as a second revival model. The essence of logistic regression is to assume that the data obeys the following logistic distribution and make parameter estimation based on maximum likelihood
Figure BDA0003419259940000101
The assumed data in the logistic regression is taken from random samples, the data in the logistic regression is derived from real pushed data, and the random sampling principle is followed in the sampling process of the original data, so that the assumed conditions of the random samples in the logistic regression can be met;
multiple collinearity also needs to be avoided in the logistic regression, the following multiple collinearity processing modes of the logistic regression model after PCA processing and the logistic regression model for the original sampling data are different, and because the vectors after PCA processing are vertically intersected pairwise, multiple collinearity does not exist, and multiple collinearity inspection does not need to be carried out; for original sampling data, multiple collinearity tests are required; in addition, the method is different from the linear regression model, the residual errors are required to be independent and distributed, and no relevant requirements are required in logic.
Maximum likelihood is one of the most widely used parameter estimation methods, based on the maximum likelihood principle. The maximum likelihood estimation method in machine learning is to find the parameter estimation with the minimum error of the whole model, i.e. to set theta (theta)0,θ1,θ2,...,θn) For a parameter vector, y is the function output and X is the function input vector, the maximum likelihood is such that the following error function.
E(θ)＝∑(y-θTX)2
This logical process is provided in SAS EG, model optimization using Fisher scoring, and model variable selection using stepwise regression (stepwise) method.
In this embodiment, the logistic regression model is trained based on the training features after the principal component analysis, the trained logistic regression model is obtained as the second survival calling model, and the survival calling model of the user is obtained by dynamically selecting the logistic modeling effect of the data subjected to the PCA dimension reduction processing, so that the precision of the survival calling model of the user is improved, and the success rate of survival calling and pushing of the user is improved.
In one embodiment, the obtaining a model with a high probability of evoking a user from the first evoking model and the second evoking model as a user evoking model comprises:
selecting AUC and ROC as evaluation indexes of the trained logistic regression model;
and respectively evaluating the first and second reviving models based on the evaluation indexes, and acquiring a model with high effect evaluation in the first and second reviving models as the user reviving model.
Specifically, for the trained logistic regression model, the AUC value is selected as the evaluation index in this embodiment. AUC values are defined as the Area under the ROC Curve (Area under the Curve of ROC (AUC ROC)), and have the following meaning: because the area is determined in 1x1 squares, the AUC must be between 0 and 1. Assuming positive above the threshold and negative below; if a positive sample and a negative sample are randomly extracted, the classifier correctly judges that the probability that the value of the positive sample is higher than that of the negative sample is equal to the AUC. The AUC value can be calculated relatively quickly by ranking the probability of whether the sample is determined to be a positive sample, and the complete calculation formula is as follows. Wherein, the sample sequence of the ith sample is represented, M and N are respectively the number of positive and negative samples,
Figure BDA0003419259940000111
is the sum of the positive sample ordering indices. The AUC formula is as follows:
Figure BDA0003419259940000121
in the case of using sample data subjected to PCA processing, the AUC value of the model is 0.717, and the effect of the model is evaluated as "normal" when the AUC value is 0.7 to 0.85 in terms of judging the classifier goodness criterion by AUC. The ROC curve is shown by an arc line in the graph, in which the horizontal axis represents negative-positive ratio (FPR) specificity and the vertical axis represents true ratio (TPR) Sensitivity, i.e., positive coverage (Sensitivity). The ROC curve is more deviated from the positive 45-degree curve as a whole, i.e., the closer to the (0,1) point, the better the model effect is, but the overall model effect is in a general level in the following figure, but the prediction of events in internet finance has a larger randomness compared with other predictions, and the results of the AUC and ROC curves above can be calculated as better results.
And respectively evaluating the first and second reviving models based on evaluation indexes, comparing the evaluation result of the first reviving model with the evaluation result of the second reviving model, and acquiring a model with high effect evaluation in the first reviving model and the second reviving model as a user reviving model.
In the embodiment, the AUC and the ROC are selected as evaluation indexes of the trained logistic regression model, the first revival model and the second revival model are evaluated based on the evaluation indexes, and the model with the high effect evaluation in the first revival model and the second revival model is obtained and used as the revival model of the user, so that the precision of the revival model of the user is improved, and the success rate of revival pushing of the user is improved.
In one embodiment, the obtaining of the information of the user to be pushed and the obtaining of the user with the high call probability based on the information of the user to be pushed and the user call model includes:
step 302, inputting the information of the users to be pushed to the user call model, and acquiring the call probability of each user to be pushed;
and step 304, screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
Specifically, fig. 3 is a schematic flowchart of a step of acquiring a user with a high call probability in an embodiment, and as shown in fig. 3, acquiring information of a user to be pushed, inputting the information of the user to be pushed into an acquired user call model for processing, acquiring call probability of each user to be pushed, screening the users to be pushed based on the call probability of each user to be pushed, comparing the call probability of each user to be pushed with a preset first threshold, and determining whether the call probability of each user to be pushed exceeds the preset first threshold; and screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
In the embodiment, the information of the users to be pushed is input into the user call model, the call probability of each user to be pushed is obtained, the users to be pushed with the call probability reaching the preset first threshold are screened out as the users with the high call probability, the users with the high call probability are obtained for information pushing through screening the user call model for the user group, and the call and push success rate of the users is improved.
In one embodiment, said pushing information to said high probability of arousal user comprises:
selecting a functional scene-based file title as an information title to be pushed, and simultaneously selecting high-welfare activities as information contents to be pushed;
and pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
Specifically, a comparison test is performed by taking the push time, the push title and the push content as comparison factors, and according to a comparison result, it can be known that the push effect of 10 points or 18 points later is not obviously different; in the aspect of pushing the titles, if the functional scene-based file is used, the pushing effect is improved, and the more the welfare is, the better the effect is; the pushing effect of the high-welfare activities is obviously better than that of the conventional function pushing activities in the aspect of pushing contents. According to the test result, when the user with high arousing probability is pushed, the functional scene-based file title is selected as the information title to be pushed, and meanwhile, high welfare activity is selected as the information content to be pushed. And pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
In the embodiment, the functional scene-based file title is selected as the information title to be pushed, the high welfare activity is selected as the information content to be pushed, information is pushed to the user with the high survival rate based on the information title to be pushed and the information content to be pushed, and the survival rate of the information title to be pushed and the information content to be pushed to the user is improved in the aspect of pushing the information content.
It should be understood that, although the steps in the flowcharts related to the embodiments as described above are sequentially displayed as indicated by arrows, the steps are not necessarily performed sequentially 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 a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be rotated or alternated with other steps or at least a part of the steps or stages in other steps.
Based on the same inventive concept, the embodiment of the application also provides an information pushing device based on the financial APP, which is used for realizing the information pushing method based on the financial APP. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the method, so that specific limitations in one or more embodiments of the information pushing device based on financial APP provided below can be referred to the limitations for the information pushing method based on financial APP, and are not described herein again.
In one embodiment, as shown in fig. 4, there is provided an information pushing apparatus 400 based on a financial APP, including: a sample obtaining module 401, a feature processing module 402, a model obtaining module 403, and an information pushing module 404, where:
a sample obtaining module 401, configured to obtain training samples, where each training sample includes multiple training features; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
a feature processing module 402, configured to perform feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
a model obtaining module 403, configured to obtain a user arousing model based on the processed training features and the logistic regression model;
an information pushing module 404, configured to obtain information of a user to be pushed, obtain a user with a high call probability based on the information of the user to be pushed and the user call model, and push information to the user with the high call probability; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In an embodiment, the model obtaining module 403 is specifically configured to: acquiring a first arousing model based on the processed training features and the logistic regression model; performing principal component analysis processing on the processed training features to obtain the training features after principal component analysis; acquiring a second revival model based on the training characteristics and the logistic regression model after the principal component analysis; and acquiring a model with high probability of calling a user from the first calling model and the second calling model as a user calling model.
In one embodiment, the model obtaining module 403 is further configured to: and training the logistic regression model based on the training characteristics after the principal component analysis, and obtaining the trained logistic regression model as a second arousing model.
In one embodiment, the model obtaining module 403 is further configured to: selecting AUC and ROC as evaluation indexes of the trained logistic regression model; and respectively evaluating the first and second reviving models based on the evaluation indexes, and acquiring a model with high effect evaluation in the first and second reviving models as the user reviving model.
In an embodiment, the information pushing module 404 is specifically configured to: inputting the information of the users to be pushed to the user call-alive model, and acquiring the call-alive probability of each user to be pushed; and screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
In one embodiment, the information pushing module 404 is further configured to: selecting a functional scene-based file title as an information title to be pushed, and simultaneously selecting high-welfare activities as information contents to be pushed; and pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
Above-mentioned information push device based on finance APP obtains the training sample, contains a plurality of training characteristics in every training sample, carries out feature processing to the training characteristic, obtains the training characteristic after the processing, based on the training characteristic after the processing and the logistic regression model, obtains the user and calls the model alive, obtains treating the propelling movement user information, based on treating propelling movement user information and user and calling the model alive, obtains the high probability of calling alive user, and to high probability of calling alive user propelling movement information generates the user through the training and calls the model alive to call out the high probability of calling alive user based on the user and carry out information push, improved and called the success rate alive to the propelling movement of user.
The modules in the financial APP-based information pushing device can be wholly or partially implemented 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 includes a non-volatile 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 financial APP-based information pushing method.
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 training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In one embodiment, the processor, when executing the computer program, further performs the steps of: acquiring a first arousing model based on the processed training features and the logistic regression model; performing principal component analysis processing on the processed training features to obtain the training features after principal component analysis; acquiring a second revival model based on the training characteristics and the logistic regression model after the principal component analysis; and acquiring a model with high probability of calling a user from the first calling model and the second calling model as a user calling model.
In one embodiment, the processor, when executing the computer program, further performs the steps of: and training the logistic regression model based on the training characteristics after the principal component analysis, and obtaining the trained logistic regression model as a second arousing model.
In one embodiment, the processor, when executing the computer program, further performs the steps of: selecting AUC and ROC as evaluation indexes of the trained logistic regression model; and respectively evaluating the first and second reviving models based on the evaluation indexes, and acquiring a model with high effect evaluation in the first and second reviving models as the user reviving model.
In one embodiment, the processor, when executing the computer program, further performs the steps of: inputting the information of the users to be pushed to the user call-alive model, and acquiring the call-alive probability of each user to be pushed; and screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
In one embodiment, the processor, when executing the computer program, further performs the steps of: selecting a functional scene-based file title as an information title to be pushed, and simultaneously selecting high-welfare activities as information contents to be pushed; and pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
The computer equipment obtains training samples, each training sample comprises a plurality of training features, the training features are subjected to feature processing, the processed training features are obtained, a user call model is obtained based on the processed training features and a logistic regression model, user information to be pushed is obtained, users with high call probability are obtained based on the user information to be pushed and the user call model, information is pushed to the users with high call probability, the user call model is generated through training, the users with high call probability are screened out based on the user call model for information pushing, and the push call success rate of the users is improved.
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 training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a first arousing model based on the processed training features and the logistic regression model; performing principal component analysis processing on the processed training features to obtain the training features after principal component analysis; acquiring a second revival model based on the training characteristics and the logistic regression model after the principal component analysis; and acquiring a model with high probability of calling a user from the first calling model and the second calling model as a user calling model.
In one embodiment, the computer program when executed by the processor further performs the steps of: and training the logistic regression model based on the training characteristics after the principal component analysis, and obtaining the trained logistic regression model as a second arousing model.
In one embodiment, the computer program when executed by the processor further performs the steps of: selecting AUC and ROC as evaluation indexes of the trained logistic regression model; and respectively evaluating the first and second reviving models based on the evaluation indexes, and acquiring a model with high effect evaluation in the first and second reviving models as the user reviving model.
In one embodiment, the computer program when executed by the processor further performs the steps of: inputting the information of the users to be pushed to the user call-alive model, and acquiring the call-alive probability of each user to be pushed; and screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
In one embodiment, the computer program when executed by the processor further performs the steps of: selecting a functional scene-based file title as an information title to be pushed, and simultaneously selecting high-welfare activities as information contents to be pushed; and pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
The storage medium obtains training samples, each training sample comprises a plurality of training features, the training features are subjected to feature processing, the processed training features are obtained, a user life calling model is obtained based on the processed training features and a logistic regression model, user information to be pushed is obtained, users with high life calling probability are obtained based on the user information to be pushed and the user life calling model, information is pushed to the users with high life calling probability, the user life calling model is generated through training, the users with high life calling probability are screened out based on the user life calling model to carry out information pushing, and the success rate of pushing life calling of the users is improved.
In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, performs the steps of:
acquiring training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a first arousing model based on the processed training features and the logistic regression model; performing principal component analysis processing on the processed training features to obtain the training features after principal component analysis; acquiring a second revival model based on the training characteristics and the logistic regression model after the principal component analysis; and acquiring a model with high probability of calling a user from the first calling model and the second calling model as a user calling model.
In one embodiment, the computer program when executed by the processor further performs the steps of: and training the logistic regression model based on the training characteristics after the principal component analysis, and obtaining the trained logistic regression model as a second arousing model.
In one embodiment, the computer program when executed by the processor further performs the steps of: selecting AUC and ROC as evaluation indexes of the trained logistic regression model; and respectively evaluating the first and second reviving models based on the evaluation indexes, and acquiring a model with high effect evaluation in the first and second reviving models as the user reviving model.
In one embodiment, the computer program when executed by the processor further performs the steps of: inputting the information of the users to be pushed to the user call-alive model, and acquiring the call-alive probability of each user to be pushed; and screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
In one embodiment, the computer program when executed by the processor further performs the steps of: selecting a functional scene-based file title as an information title to be pushed, and simultaneously selecting high-welfare activities as information contents to be pushed; and pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
The computer program product obtains training samples, each training sample comprises a plurality of training features, the training features are subjected to feature processing, the processed training features are obtained, a user call model is obtained based on the processed training features and a logistic regression model, user information to be pushed is obtained, users with high call probability are obtained based on the user information to be pushed and the user call model, information is pushed to the users with high call probability, the user call model is generated through training, the users with high call probability are screened out based on the user call model to carry out information pushing, and the push call success rate of the users is improved.
It should be noted that, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party.
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, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high-density embedded nonvolatile Memory, resistive Random Access Memory (ReRAM), Magnetic Random Access Memory (MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (PCM), graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. 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 databases referred to in various embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing based data processing logic devices, etc., without limitation.
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 present application. 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 application shall be subject to the appended claims.

Claims (10)

1. An information pushing method based on financial APP is characterized by comprising the following steps:
acquiring training samples, wherein each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
performing feature processing on the training features to obtain processed training features; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
acquiring a user calling model based on the processed training characteristics and the logistic regression model;
acquiring user information to be pushed, acquiring users with high probability of call based on the user information to be pushed and the user call model, and pushing information to the users with high probability of call; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
2. The method of claim 1, wherein obtaining a user arousal model based on the processed training features and a logistic regression model comprises:
acquiring a first arousing model based on the processed training features and the logistic regression model;
performing principal component analysis processing on the processed training features to obtain the training features after principal component analysis;
acquiring a second revival model based on the training characteristics and the logistic regression model after the principal component analysis;
and acquiring a model with high probability of calling a user from the first calling model and the second calling model as a user calling model.
3. The method of claim 2, wherein obtaining a second evoking model based on the principal component analyzed training features and a logistic regression model comprises:
and training the logistic regression model based on the training characteristics after the principal component analysis, and obtaining the trained logistic regression model as a second arousing model.
4. The method of claim 2, wherein the obtaining a model with a high probability of evoking a user from the first and second evoking models as a user evoking model comprises:
selecting AUC and ROC as evaluation indexes of the trained logistic regression model;
and respectively evaluating the first and second reviving models based on the evaluation indexes, and acquiring a model with high effect evaluation in the first and second reviving models as the user reviving model.
5. The method of claim 1, wherein the obtaining of the user with high arousal probability based on the information of the user to be pushed and the user arousal model comprises:
inputting the information of the users to be pushed to the user call-alive model, and acquiring the call-alive probability of each user to be pushed;
and screening out users to be pushed, the survival probability of which reaches a preset first threshold value, as the users with high survival probability.
6. The method of claim 1, wherein said pushing information to said high probability of arousal user comprises:
selecting a functional scene-based file title as an information title to be pushed, and simultaneously selecting high-welfare activities as information contents to be pushed;
and pushing information to the user with high probability of arousing based on the information title to be pushed and the information content to be pushed.
7. An information push device based on finance APP, its characterized in that, the device includes:
the device comprises a sample acquisition module, a data acquisition module and a data processing module, wherein the sample acquisition module is used for acquiring training samples, and each training sample comprises a plurality of training characteristics; the training characteristics comprise at least one of user registration information, user asset information, service opening information, key function use information, region click information, user transaction information and internet product use information;
the characteristic processing module is used for carrying out characteristic processing on the training characteristics to obtain the processed training characteristics; the characteristic processing process comprises at least one of missing value processing, variable derivation, data standardization and classification variable processing;
the model acquisition module is used for acquiring a user calling model based on the processed training characteristics and the logistic regression model;
the information pushing module is used for acquiring user information to be pushed, acquiring a user with a high call probability based on the user information to be pushed and the user call model, and pushing information to the user with the high call probability; and the information type in the user information to be pushed is the same as the information type in the training characteristics.
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.
10. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 6 when executed by a processor.
CN202111557181.4A 2021-12-18 2021-12-18 Information pushing method and device based on financial APP and computer equipment Pending CN114238763A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111557181.4A CN114238763A (en) 2021-12-18 2021-12-18 Information pushing method and device based on financial APP and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111557181.4A CN114238763A (en) 2021-12-18 2021-12-18 Information pushing method and device based on financial APP and computer equipment

Publications (1)

Publication Number Publication Date
CN114238763A true CN114238763A (en) 2022-03-25

Family

ID=80758731

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111557181.4A Pending CN114238763A (en) 2021-12-18 2021-12-18 Information pushing method and device based on financial APP and computer equipment

Country Status (1)

Country Link
CN (1) CN114238763A (en)

Similar Documents

Publication Publication Date Title
Yin et al. IGRF-RFE: a hybrid feature selection method for MLP-based network intrusion detection on UNSW-NB15 dataset
CN111724083A (en) Training method and device for financial risk recognition model, computer equipment and medium
US11693917B2 (en) Computational model optimizations
US11875408B2 (en) Techniques for accurate evaluation of a financial portfolio
Hung et al. Customer segmentation using hierarchical agglomerative clustering
US20220253856A1 (en) System and method for machine learning based detection of fraud
Nalić et al. Importance of data pre-processing in credit scoring models based on data mining approaches
CN115983900A (en) Method, apparatus, device, medium, and program product for constructing user marketing strategy
Kang et al. A CWGAN-GP-based multi-task learning model for consumer credit scoring
CN113408582B (en) Training method and device for feature evaluation model
Tounsi et al. Credit scoring in the age of big data–A state-of-the-art
US20220414262A1 (en) Rule-based anonymization of datasets
CN112927719B (en) Risk information evaluation method, apparatus, device and storage medium
CN114170000A (en) Credit card user risk category identification method, device, computer equipment and medium
CN114693409A (en) Product matching method, device, computer equipment, storage medium and program product
CN114238763A (en) Information pushing method and device based on financial APP and computer equipment
CN114529399A (en) User data processing method, device, computer equipment and storage medium
CN114049204A (en) Suspicious transaction data entry method, device, computer equipment and computer-readable storage medium
Slabchenko et al. Development of models for imputation of data from social networks on the basis of an extended matrix of attributes
CN112884028A (en) System resource adjusting method, device and equipment
Zhuang et al. A deep metric learning approach for weakly supervised loan default prediction
Liu et al. A comparison of machine learning algorithms for prediction of past due service in commercial credit
Zhao An evolutionary intelligent data analysis in promoting smart community
Huang et al. Variable selection in classification model via quadratic programming
CN114281994B (en) Text clustering integration method and system based on three-layer weighting model

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