CN108596393A - A kind of training method and device of Electronic Finance resources model - Google Patents
A kind of training method and device of Electronic Finance resources model Download PDFInfo
- Publication number
- CN108596393A CN108596393A CN201810398234.4A CN201810398234A CN108596393A CN 108596393 A CN108596393 A CN 108596393A CN 201810398234 A CN201810398234 A CN 201810398234A CN 108596393 A CN108596393 A CN 108596393A
- Authority
- CN
- China
- Prior art keywords
- electronic finance
- model
- characteristic value
- value
- resource
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
Abstract
This application provides a kind of training method and device of Electronic Finance resources model, this method includes:The history media data of publisher based on Electronic Finance resource determines the medium property feature of the Electronic Finance resource;Based on the historical trading data of the Electronic Finance resource, the transaction attributive character of the Electronic Finance resource is determined;Screening Treatment is carried out to the medium property feature and the transaction attributive character, obtains the key feature for influencing Electronic Finance resource fluctuations;It builds at least two preset models using the fluctuation characteristic for characterizing the Electronic Finance resource fluctuations situation as dependent variable using the key feature as independent variable and is trained;Based on the prediction result that at least two preset model obtains, fusion treatment is carried out to described two preset models using preset model fusion method, obtains Electronic Finance resources model.
Description
Technical field
This application involves data analysis technique fields, in particular to a kind of instruction of Electronic Finance resources model
Practice method and apparatus.
Background technology
Securities market is the basis of national economic development Capital Flow.The evolutionary process of security price is by many economical
Body and economic factor participate in co-determination jointly, cause the model of volatility complex, it is difficult to effectively be predicted.
General Prediction of Stock Index method is carries out the prediction of stock using the financial information of company, the historical information of stock, due to right
The factor that the fluctuation of stock has an impact is more, and therefore, the effect of existing prediction technique prediction is simultaneously bad.
Invention content
In view of this, the application's is designed to provide a kind of training method and dress of Electronic Finance resources model
It sets, for solving the problems, such as that Electronic Finance resources accuracy is low in the prior art.
In a first aspect, the embodiment of the present application provides a kind of training method of Electronic Finance resources model, this method
Including:
The history media data of publisher based on Electronic Finance resource determines the medium property of the Electronic Finance resource
Feature;
Based on the historical trading data of the Electronic Finance resource, determine that the transaction attribute of the Electronic Finance resource is special
Sign;
Screening Treatment is carried out to the medium property feature and the transaction attributive character, obtains influencing Electronic Finance resource
The key feature of fluctuation;
Using the key feature as independent variable, will characterize the fluctuation characteristic of the Electronic Finance resource fluctuations situation as
Dependent variable builds at least two preset models and is trained;
Based on the prediction result that at least two preset model obtains, using preset model fusion method to described two
Preset model carries out fusion treatment, obtains Electronic Finance resources model.
Optionally, the prediction result obtained based at least two preset model, using preset model fusion method pair
At least two preset model carries out fusion treatment, specifically includes:
Using the prediction result of at least two preset model as independent variable, using the fluctuation characteristic as dependent variable,
Structure Fusion Model is simultaneously trained.
Optionally, described using key feature as independent variable, using the fluctuation characteristic as dependent variable, structure at least two
Preset model is simultaneously trained, and is specifically included:
Based on the history media data and historical trading data, the characteristic value of the key feature is determined;
Based on the historical volatility data, the characteristic value of fluctuation characteristic is determined;Wherein, the characteristic value pair of the fluctuation characteristic
The historical time answered compares the corresponding historical time lag of characteristic value of the key feature;
Determine at least two preset models;And
For each preset model, using the characteristic value of the key feature as the value of independent variable, by corresponding fluctuation spy
Value of the characteristic value of sign as dependent variable, is trained the preset model, obtains at least two preset models for completing training.
Optionally, at least two preset model includes:Neural network prediction model;
For each preset model, using the characteristic value of key feature as the value of independent variable, by the spy of the fluctuation characteristic
Value of the value indicative as dependent variable, is trained the preset model, specifically includes:
For multiple neural network models, following training operation is executed respectively, wherein multiple neural networks have different
The neural network number of plies:
Using the characteristic value of key feature as the value of independent variable, using the characteristic value of the fluctuation characteristic as dependent variable
Value, is trained Current Situation of Neural Network model, obtains the index value of the pre-set level for weighing model prediction accuracy;
Using the highest neural network model of index value as finally determining neural network prediction model.
Optionally, the described pair of determining medium property feature and the transaction attributive character carry out Screening Treatment, obtain
To the key feature for influencing Electronic Finance resource fluctuations, including:
The medium property feature and the transaction attributive character degree of correlation between the fluctuation characteristic respectively are calculated,
It is more than that the medium property feature of default relevance threshold and/or transaction attributive character are determined as the first effect characteristics by the degree of correlation
Collection;
The medium property feature and the transaction attributive character are carried out at screening using decision tree Variable Selection method
Reason, obtains the second effect characteristics collection;
The first effect characteristics collection and the second effect characteristics collection are carried out using preset characteristic value Processing Algorithm
Processing, obtains third effect characteristics collection;
The expansion factor value that third effect characteristics concentrate each third effect characteristics is calculated, by expansion factor value less than default
The third effect characteristics of expansion factor threshold value are determined as the key feature.
Optionally, described that the first effect characteristics collection and described second are influenced using preset characteristic value Processing Algorithm
Feature set is handled, and third effect characteristics collection is obtained, including:
The history of the history media data of publisher based on the Electronic Finance resource and the Electronic Finance resource is handed over
It is each to determine that first effect characteristics concentrate the characteristic value of each first effect characteristics and second effect characteristics to concentrate for easy data
The characteristic value of second effect characteristics;
The characteristic value of each first effect characteristics and the characteristic value of each second effect characteristics are become as oneself
The value of amount inputs the characteristic value that training is completed in advance and handles model, obtains each first effect characteristics and each second effect characteristics
Significance;
Significance is more than corresponding first effect characteristics of significance threshold value or the second effect characteristics are determined as the third
Effect characteristics collection.
Second aspect, the embodiment of the present application provide a kind of Electronic Finance resource trends prediction technique, and this method includes:
The history media data of publisher based on Electronic Finance resource to be predicted determines the Electronic Finance money to be predicted
The characteristic value of the medium property feature in source;And
Based on the historical trading data of the Electronic Finance resource to be predicted, the Electronic Finance resource to be predicted is determined
The characteristic value for attributive character of merchandising;
Using the characteristic value of the characteristic value of determining medium property feature and transaction attributive character as the value of independent variable, input
The Electronic Finance resources model of above-mentioned determination, predicts the rise probability of the Electronic Finance resource.
The third aspect, the embodiment of the present application provide a kind of training device of Electronic Finance resources model, the device
Including:
First determining module is used for the history media data of the publisher based on Electronic Finance resource, determines the finance
The medium property feature of e-sourcing;
Second determining module is used for the historical trading data based on the Electronic Finance resource, determines the Electronic Finance
The transaction attributive character of resource;
Screening module obtains shadow for carrying out Screening Treatment to the medium property feature and the transaction attributive character
Ring the key feature of Electronic Finance resource fluctuations;
Module is built, for using the key feature as independent variable, the Electronic Finance resource fluctuations situation will to be characterized
Fluctuation characteristic as dependent variable, at least two preset models of structure are simultaneously trained;
Processing module, the prediction result for being obtained based at least two preset model, is merged using preset model
Method carries out fusion treatment to described two preset models, obtains Electronic Finance resources model.
Fourth aspect, the embodiment of the present application provide a kind of Electronic Finance resource trends prediction meanss, which includes:
First determining module is used for the history media data of the publisher based on Electronic Finance resource to be predicted, determines institute
State the characteristic value of the medium property feature of Electronic Finance resource to be predicted;
Second determining module is used for the historical trading data based on the Electronic Finance resource to be predicted, is waited for described in determination
Predict the characteristic value of the transaction attributive character of Electronic Finance resource
Prediction module, the characteristic value of the medium property feature for will determine and the characteristic value of transaction attributive character are as certainly
The value of variable inputs the Electronic Finance resources model of above-mentioned determination, predicts the rise probability of the Electronic Finance resource.
5th aspect, the embodiment of the present application provide a kind of computer equipment and include memory, processor and be stored in institute
The computer program that can be run on memory and on the processor is stated, the processor executes real when the computer program
The step of showing above-mentioned method.
The training method of Electronic Finance resources model provided by the embodiments of the present application is issued according to Electronic Finance resource
The history media data of side, determines the medium property feature of Electronic Finance resource, the historical trading number based on Electronic Finance resource
According to determining transaction attributive character carries out Screening Treatment to medium property feature and transaction attributive character, obtains influence Electronic Finance
The key feature of resource fluctuations builds at least two preset models and is trained, be based on according to key feature and fluctuation characteristic
The prediction result that at least two preset models obtain is carried out using two preset models of preset model fusion method pair at fusion
Reason, obtaining can be to Electronic Finance resources model that Electronic Finance resource trends are predicted.When building model, pass through
History media data and historical trading data are handled, increased on the influential feature of Electronic Finance resource tool so that
It is more diversified to build the independent variable that model needs, influence of the comprehensive considering various effects to Electronic Finance resource fluctuations, effectively
Improve the accuracy of Electronic Finance resources.
To enable the above objects, features, and advantages of the application to be clearer and more comprehensible, preferred embodiment cited below particularly, and coordinate
Appended attached drawing, is described in detail below.
Description of the drawings
It, below will be to needed in the embodiment attached in order to illustrate more clearly of the technical solution of the embodiment of the present application
Figure is briefly described, it should be understood that the following drawings illustrates only some embodiments of the application, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 is a kind of flow signal of training method of Electronic Finance resources model provided by the embodiments of the present application
Figure;
Fig. 2 is a kind of flow diagram of Electronic Finance resource trends prediction technique provided by the embodiments of the present application;
Fig. 3 is a kind of structural representation of the training device of Electronic Finance resources model provided by the embodiments of the present application
Figure;
Fig. 4 is a kind of structural schematic diagram of Electronic Finance resource trends prediction meanss provided by the embodiments of the present application;
Fig. 5 is a kind of structural schematic diagram of computer equipment 500 provided by the embodiments of the present application;
Fig. 6 is a kind of structural schematic diagram of computer equipment 600 provided by the embodiments of the present application.
Specific implementation mode
To keep the purpose, technical scheme and advantage of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
Middle attached drawing, technical solutions in the embodiments of the present application are clearly and completely described, it is clear that described embodiment is only
It is some embodiments of the present application, instead of all the embodiments.The application being usually described and illustrated herein in the accompanying drawings is real
Applying the component of example can be arranged and designed with a variety of different configurations.Therefore, below to the application's for providing in the accompanying drawings
The detailed description of embodiment is not intended to limit claimed scope of the present application, but is merely representative of the selected reality of the application
Apply example.Based on embodiments herein, institute that those skilled in the art are obtained without making creative work
There is other embodiment, shall fall in the protection scope of this application.
The embodiment of the present application provides a kind of training method of Electronic Finance resources model, as shown in Figure 1, this method
Including:
S101, the history media data of the publisher based on Electronic Finance resource determine the matchmaker of the Electronic Finance resource
Body attributive character;
Here, Electronic Finance resource can be but not limited to stock, security, fund etc., the publisher of Electronic Finance resource
Enterprise, company for generally issuing shares etc.;History media data is obtained from default platform, and default platform can be today
Top news, Netease's news, Sina News, Sina weibo etc., history media data can be new for financial and economic news, INDUSTRY OVERVIEW, public opinion
News, policy news etc.;Medium property feature may include having industry label, main management label, leadership's label, rival's mark
Label, upstream and downstream label, news quantity, average review quantity, number of reviews maximum value etc., wherein industry label, main management label,
Leadership's label, rival's label, upstream and downstream label are usually to be determined according to the information of the publisher of Electronic Finance resource,
News data, average review quantity, number of reviews maximum value are generally based on the determination of history media data.
S102 determines the transaction category of the Electronic Finance resource based on the historical trading data of the Electronic Finance resource
Property feature;
Here, historical trading data is generally stock certificate data, and stock certificate data generally comprises the closing price of stock, highest
Valence, lowest price, trading volume, turnover rate etc., transaction attributive character generally comprise transaction value, turnover rate etc., the application to this not
Give limitation.
S103 carries out Screening Treatment to the medium property feature and the transaction attributive character, obtains influencing financial electricity
The key feature of child resource fluctuation;
Here, key feature is generally the feature bigger to the influence of fluctuations of Electronic Finance resource.
Screening Treatment is carried out to the determining medium property feature and the transaction attributive character, obtains influencing financial electricity
The key feature of child resource fluctuation, specifically includes following steps:
The medium property feature and the transaction attributive character degree of correlation between the fluctuation characteristic respectively are calculated,
It is more than that the medium property feature of default relevance threshold and/or transaction attributive character are determined as the first effect characteristics by the degree of correlation
Collection;
The medium property feature and the transaction attributive character are carried out at screening using decision tree Variable Selection method
Reason, obtains the second effect characteristics collection;
The first effect characteristics collection and the second effect characteristics collection are carried out using preset characteristic value Processing Algorithm
Processing, obtains third effect characteristics collection;
The expansion factor value that third effect characteristics concentrate each third effect characteristics is calculated, by expansion factor value less than default
The third effect characteristics of expansion factor threshold value are determined as the key feature.
It is calculated by the following formula medium property feature or transaction the attributive character degree of correlation between fluctuation characteristic respectively
r:
Wherein, r is the degree of correlation of medium property feature or merchandise attributive character and fluctuation characteristic, and x is medium property spy
Sign or transaction attributive character, y are fluctuation characteristic, and N is the number of medium property feature or the corresponding characteristic value of attributive character of merchandising,
Generally positive integer.
It is illustrated for calculating the degree of correlation between medium property feature and fluctuation characteristic, medium property feature x is
The number N of the corresponding characteristic value of a, a is 3, and a is corresponding with the feature of fluctuation characteristic y in the index value of each historical time section
Value, that is, 3 values of a are corresponding with the value of 3 fluctuation characteristic y, will be in each characteristic value of a and the input of the value of corresponding fluctuation characteristic y
Relatedness computation formula is stated, and then obtains the degree of correlation between medium property feature a and corresponding fluctuation characteristic.Wherein, history
Period can be 1 day, 1 week, 1 month, 1 season etc., the application not limits this.Other medium property features
And transaction attributive character and the calculating process of the degree of correlation of fluctuation characteristic it is identical as above-mentioned example, no longer solved one by one herein
It releases.
Can will be more than pre- after obtaining the degree of correlation between medium property feature, transaction attributive character and fluctuation characteristic
If relevance threshold medium property feature or transaction attributive character be determined as the first effect characteristics collection, can also be according to correlation
The sequence of degree from high to low, is ranked up each degree of correlation, quantity threshold media category before being selected from the degree of correlation after sequence
Property feature and/or transaction attributive character be determined as the first effect characteristics collection.Finally obtained first effect characteristics concentration may be only
Including medium property feature, it is also possible to only include transaction attributive character, it is also possible to not only include medium property feature but also including transaction
Attributive character, should be subject to actual conditions.
The method for carrying out Screening Treatment to medium property feature and transaction attributive character includes but not limited to decision tree variable
Screening technique, decision variable screening technique have detailed introduction, are no longer excessively illustrated herein in the prior art.
The first effect characteristics collection and the second effect characteristics collection are carried out using preset characteristic value Processing Algorithm
Processing, obtains third effect characteristics collection, specifically includes following steps:
The history of the history media data of publisher based on the Electronic Finance resource and the Electronic Finance resource is handed over
It is each to determine that first effect characteristics concentrate the characteristic value of each first effect characteristics and second effect characteristics to concentrate for easy data
The characteristic value of second effect characteristics;
The characteristic value of each first effect characteristics and the characteristic value of each second effect characteristics are become as oneself
The value of amount inputs the characteristic value that training is completed in advance and handles model, obtains each first effect characteristics and each second effect characteristics
Significance;
Significance is more than corresponding first effect characteristics of significance threshold value or the second effect characteristics are determined as the third
Effect characteristics collection.
Here, characteristic value processing model can be Logic Regression Models, neural network model etc., and training characteristics value handles mould
The method of type has detailed introduction in the prior art, is no longer excessively illustrated herein.
In specific implementation, the first effect characteristics that the first effect characteristics are concentrated may be that medium property feature may be to hand over
Easy attributive character, the second effect characteristics that the second effect characteristics are concentrated may be that medium property feature may be that transaction attribute is special
Sign, according to the history media data of the publisher of Electronic Finance resource, it may be determined that the characteristic value of medium property feature, according to gold
Melt the historical trading data of e-sourcing, it may be determined that the characteristic value for attributive character of merchandising.
For example, history media data can be that certain issuance of shares side is new in the financial and economic news of multiple historical time sections, industry
It hears, public opinion news etc., the news quantity of each historical time section, the number of reviews of every news is counted, for each history
Period, the number of reviews of each news of counting statistics and value average value, using the average value as the historical time section
Average review quantity, and record the maximum value of the number of reviews of the historical time section news.
After obtaining history media data, statistics includes the industry label of the issuance of shares side or manages label mainly or lead
The quantity of the news of conducting shell label etc. will include that the quantity of news of corresponding label is determined as the spy of corresponding medium property feature
Value indicative, e.g., the industry belonging to the issuance of shares side are the communications industry, include communication-related information in 4 news, then media category
Property feature in industry label characteristic value be 4.Above-mentioned example can be referred in the characteristic value for calculating each medium property feature.
In the historical trading data for getting Electronic Finance resource, e.g., closing price, the highest price of each historical time section
Deng determining the characteristic value of closing price in each historical time section transaction attributive character.
It is defeated using the characteristic value of each first effect characteristics and the characteristic value of each second effect characteristics as the value of independent variable
After entering the characteristic value processing model that training is completed in advance, the significance of each first effect characteristics and each second effect characteristics are obtained
Significance is more than the first effect characteristics of significance threshold value and the second effect characteristics is ultimately determined to third and influences spy by significance
Collection.Wherein, it may be transaction attributive character that the second effect characteristics that third effect characteristics are concentrated, which may be medium property feature,
Significance is usually to handle what model obtained by characteristic value, and the prior art has detailed introduction, is not carrying out herein
It speaks more bright.
After obtaining third effect characteristics collection, third feature is calculated according to following formula, any one third is concentrated to influence spy
The expansion factor VIF of sign:
VIF=1/ (1-r2)
Wherein, VIF is the expansion factor of any one third effect characteristics;R be third effect characteristics and fluctuation characteristic it
Between the degree of correlation.
After the expansion factor value for obtaining each third effect characteristics, the third less than default expansion factor value is influenced
Feature is determined as key feature.Sequence that can also be according to expansion factor value from low to high, is ranked up each expansion factor value,
Quantity threshold third effect characteristics are set to key feature before being selected from the expansion factor value after sequence.
S104, it is using the key feature as independent variable, the fluctuation for characterizing the Electronic Finance resource fluctuations situation is special
Sign is used as dependent variable, builds at least two preset models and is trained;
Using key feature as independent variable, using the fluctuation characteristic as dependent variable, at least two preset models of structure are simultaneously
It is trained, specifically includes:
Based on the history media data and historical trading data, the characteristic value of the key feature is determined;
Based on the historical volatility data, the characteristic value of fluctuation characteristic is determined;Wherein, the characteristic value pair of the fluctuation characteristic
The historical time answered compares the corresponding historical time lag of characteristic value of the key feature;
Determine at least two preset models;And
For each preset model, using the characteristic value of the key feature as the value of independent variable, by corresponding fluctuation spy
Value of the characteristic value of sign as dependent variable, is trained the preset model, obtains at least two preset models for completing training.
Here, in the characteristic value of the characteristic value and fluctuation characteristic that determine key feature, the history media data of acquisition and
Historical trading data is generally the data of same historical time, and historical volatility data lag behind history media data or historical trading
The time of data, for example, the history media data and historical trading data that obtain are the data of the first historical time section, acquisition
Historical volatility data are the data of the second historical time section, and the second historical time section lags behind the first historical time section, and second goes through
The history period can be next period of the first historical time section, can also be intersegmental every present count with the first historical time
The mesh period;It, will be determining when two preset models are respectively logistic regression prediction model and neural network prediction model
Value of the characteristic value of key feature as independent variable, using the characteristic value of fluctuation characteristic as the value of dependent variable, difference input logic
Regressive prediction model and neural network prediction model are trained, and obtain the logistic regression prediction model and nerve net of completing training
Network prediction model.
The principle of logistic regression prediction model is as follows:
Multivariate linear model is:H (x)=a0+a1x1+a2x2+…+anxn
Wherein, h (x) is the dependent variable of multivariate linear model, a0、a1、……anFor the weight of independent variable, x1、x2、……xn
For the independent variable of multivariate linear model.
Classified to article using multivariate linear model, presets threshold values, it is then that all dependent variable h (x) are big
It is divided into one kind in the sample of threshold values, others are divided into another kind of.But this mode has a problem that, since the value of h (x) is to appoint
Size of anticipating, the selection of threshold values is that a difficult thing is normalized it for the ease of the selection of threshold values.
If threshold values is:T, then
Wherein, ha(x) it is to utilize multivariate linear model prediction result;
Assuming that:
a0=a0- t, aTX=a0+a1x1+…+anxn
Using S types (sigmoid) function pair herein, it is normalized.
If at this point, estimate parameter using square smallest error function, since the function after normalization is non-convex function, therefore
And its minimum value cannot be found using gradient descent method.But estimate model parameter using the method for Maximum-likelihood estimation.
Due to being two classification, it can be assumed that:
P (y=1 | xi)=ha(xi), p (y=0 | xi)=1-ha(xi)
Wherein, P (y=1 | xi) it is the probability that prediction result is 1;
ha(xi) be
So likelihood function is:
Wherein, h (xi) be
M is positive integer;
Log-likelihood function L (a):
To L (a) maximizings, the estimated value of a is obtained.
When preset model is neural network prediction model, since neural network has the different neural network numbers of plies, needle
To multiple neural network models, following training operation is executed respectively:
Using the characteristic value of key feature as the value of independent variable, using the characteristic value of the fluctuation characteristic as dependent variable
Value, is trained Current Situation of Neural Network model, obtains the index value of the pre-set level for weighing model prediction accuracy;
Using the highest neural network model of index value as finally determining neural network prediction model.
In specific implementation, the number of plies of neural network could be provided as most 10 layers, and since one layer, often one layer of increase is right
Answer a neural network model, using the characteristic value of key feature as the value of independent variable, using the characteristic value of fluctuation characteristic as because
The value of variable inputs above-mentioned neural network model and is trained respectively, obtains each neural network model parameter, and weigh model
The index value of forecasting accuracy, e.g., the index value can be that Andrei Kolmogorov-Si meter Nuo Fu examines (Kolmogorov-
Smirnov test, KS) check value, area (the Area under Curve of below recipient's operating characteristic curve
Receiver operating characteristic curve, AUC) index value.In practical applications, it can obtain simultaneously
KS check values and AUC index values.
The accuracy of the bigger characterization model prediction of index value is higher, therefore, by the corresponding neural network of highest index value
Model is as finally determining neural network prediction model, and e.g., when the neural network number of plies is 9, corresponding index value is maximum, the god
It is final neural network prediction model through 9 corresponding neural network model of the network number of plies.Wherein, highest index value can be
KS check values, can also be AUC index values, and the application not limits this.
S105, based on the prediction result that at least two preset model obtains, using preset model fusion method to institute
It states two preset models and carries out fusion treatment, obtain Electronic Finance resources model.
In the prediction result obtained based at least two preset model, using preset model fusion method to it is described extremely
Few two preset models carry out fusion treatment, specifically include following steps:
Using the prediction result of at least two preset model as independent variable, using the fluctuation characteristic as dependent variable,
Structure Fusion Model is simultaneously trained.
Here, fusion treatment is carried out to preset model, finally obtains Electronic Finance resources model, can obtained more preferable
Prediction goodness, increase Electronic Finance resources accuracy.Model Fusion has following two methods:
Method one:Model accumulates (Model stacking), votes the prediction result of each preset model, is taken using minority
From most principles, usually the prediction result of several preset models being weighted and is averaging, weights are directly proportional to model prediction goodness,
It is inversely proportional with the uncertainty of model.
Method two:Model integrated (Model ensemble), using the prediction result of each preset model as output valve, instruction
Practice a new grader, then, using the prediction result of trained grader as Electronic Finance resources model most
Whole prediction result.
In specific implementation, two preset models are respectively logistic regression prediction model and neural network prediction model, are incited somebody to action
Value of the characteristic value of the corresponding key feature of first historical time section as independent variable, by the fluctuation characteristic of the second historical time section
Value of the characteristic value as dependent variable, respectively input logic regressive prediction model obtain the first prediction result, input final determine
Neural network prediction model obtain the second prediction result.Wherein, the second historical time section generally lags behind the first historical time
Section.
Further, using the first prediction result and the second prediction result as the value of independent variable, by the second historical time section
Value of the characteristic value of fluctuation characteristic as dependent variable, the Fusion Model for inputting structure are trained, and finally obtain Electronic Finance money
Source prediction model, so that user predicts the up-trend of Electronic Finance resource using the Electronic Finance prediction model.
The embodiment of the present application provides a kind of Electronic Finance resource trends prediction technique, as shown in Fig. 2, this method include with
Lower step:
S201, the history media data of the publisher based on Electronic Finance resource to be predicted determine the finance to be predicted
The characteristic value of the medium property feature of e-sourcing;And
Here, in the training method of the publisher of Electronic Finance resource to be predicted and above-mentioned Electronic Finance resources model
The publisher of Electronic Finance resource be identical publisher, determine that Electronic Finance to be predicted provides herein according to history media data
The method of the characteristic value of the medium property feature in source, it is identical as method in the training method of Electronic Finance resources model, this
Place is no longer excessively illustrated.
S202 determines the Electronic Finance to be predicted based on the historical trading data of the Electronic Finance resource to be predicted
The characteristic value of the transaction attributive character of resource;
Here, the characteristic value of the transaction attributive character of Electronic Finance resource to be predicted is determined herein according to historical trading data
Method, it is identical as method in the training method of Electronic Finance resources model, no longer excessively illustrated herein.
S203, using the characteristic value of the characteristic value of determining medium property feature and transaction attributive character as independent variable
Value, inputs the Electronic Finance resources model of above-mentioned determination, predicts the rise probability of the Electronic Finance resource.
The embodiment of the present application provides a kind of training device of Electronic Finance resources model, as shown in figure 3, the device
Including:
First determining module 31 is used for the history media data of the publisher based on Electronic Finance resource, determines the gold
Melt the medium property feature of e-sourcing;
Second determining module 32 is used for the historical trading data based on the Electronic Finance resource, determines the finance electricity
The transaction attributive character of child resource;
Screening module 33 is obtained for carrying out Screening Treatment to the medium property feature and the transaction attributive character
Influence the key feature of Electronic Finance resource fluctuations;
Module 34 is built, for using the key feature as independent variable, the Electronic Finance resource fluctuations feelings will to be characterized
The fluctuation characteristic of condition builds at least two preset models and is trained as dependent variable;
Processing module 35, the prediction result for being obtained based at least two preset model, is melted using preset model
Conjunction method carries out fusion treatment to described two preset models, obtains Electronic Finance resources model.
Optionally, processing module 35 is specifically used for:
Using the prediction result of at least two preset model as independent variable, using the fluctuation characteristic as dependent variable,
Structure Fusion Model is simultaneously trained.
Optionally, structure module 34 is specifically used for:
Based on the history media data and historical trading data, the characteristic value of the key feature is determined;
Based on the historical volatility data, the characteristic value of fluctuation characteristic is determined;Wherein, the characteristic value pair of the fluctuation characteristic
The historical time answered compares the corresponding historical time lag of characteristic value of the key feature;
Determine at least two preset models;And
For each preset model, using the characteristic value of the key feature as the value of independent variable, by corresponding fluctuation spy
Value of the characteristic value of sign as dependent variable, is trained the preset model, obtains at least two preset models for completing training.
Optionally, at least two preset model includes:Neural network prediction model;
Structure module 34 is additionally operable to:
For multiple neural network models, following training operation is executed respectively, wherein multiple neural networks have different
The neural network number of plies:
Using the characteristic value of key feature as the value of independent variable, using the characteristic value of the fluctuation characteristic as dependent variable
Value, is trained Current Situation of Neural Network model, obtains the index value of the pre-set level for weighing model prediction accuracy;
Using the highest neural network model of index value as finally determining neural network prediction model.
Optionally, screening module 33 is specifically used for:
The medium property feature and the transaction attributive character degree of correlation between the fluctuation characteristic respectively are calculated,
It is more than that the medium property feature of default relevance threshold and/or transaction attributive character are determined as the first effect characteristics by the degree of correlation
Collection;
The medium property feature and the transaction attributive character are carried out at screening using decision tree Variable Selection method
Reason, obtains the second effect characteristics collection;
The first effect characteristics collection and the second effect characteristics collection are carried out using preset characteristic value Processing Algorithm
Processing, obtains third effect characteristics collection;
The expansion factor value that third effect characteristics concentrate each third effect characteristics is calculated, by expansion factor value less than default
The third effect characteristics of expansion factor threshold value are determined as the key feature.
Optionally, screening module 33 is additionally operable to:
The history of the history media data of publisher based on the Electronic Finance resource and the Electronic Finance resource is handed over
It is each to determine that first effect characteristics concentrate the characteristic value of each first effect characteristics and second effect characteristics to concentrate for easy data
The characteristic value of second effect characteristics;
The characteristic value of each first effect characteristics and the characteristic value of each second effect characteristics are become as oneself
The value of amount inputs the characteristic value that training is completed in advance and handles model, obtains each first effect characteristics and each second effect characteristics
Significance;
Significance is more than corresponding first effect characteristics of significance threshold value or the second effect characteristics are determined as the third
Effect characteristics collection.
The embodiment of the present application provides a kind of Electronic Finance resource trends prediction meanss, as shown in figure 4, the device includes:
First determining module 41 is used for the history media data of the publisher based on Electronic Finance resource to be predicted, determines
The characteristic value of the medium property feature of the Electronic Finance resource to be predicted;
Second determining module 42, be used for the historical trading data based on the Electronic Finance resource to be predicted, determine described in
The characteristic value of the transaction attributive character of Electronic Finance resource to be predicted
The characteristic value of prediction module 43, the characteristic value of the medium property feature for will determine and transaction attributive character as
The value of independent variable, the Electronic Finance resources model determined in input, predicts the rise probability of the Electronic Finance resource.
Corresponding to the training method of the Electronic Finance resources model in Fig. 1, the embodiment of the present application also provides one kind
Computer equipment 500, as shown in figure 5, the equipment includes memory 501, processor 502 and is stored on the memory 501 simultaneously
The computer program that can be run on the processor 502, wherein above-mentioned processor 502 is realized when executing above computer program
The training method of above-mentioned Electronic Finance resources model.
Specifically, above-mentioned memory 501 and processor 502 can be general memory and processor, do not do have here
Body limits, and when the computer program of 502 run memory 501 of processor storage, it is pre- to be able to carry out above-mentioned Electronic Finance resource
The training method of model is surveyed, to solve the problems, such as that prior art Electronic Finance resources accuracy is low, the application is being built
When model, by handling history media data and historical trading data, increasing to have Electronic Finance resource influences
Feature so that structure model need independent variable it is more diversified, comprehensive considering various effects are to Electronic Finance resource fluctuations
Influence, effectively improve the accuracy of Electronic Finance resources.
Corresponding to the training method of the Electronic Finance resources model in Fig. 1, the embodiment of the present application also provides one kind
Computer readable storage medium is stored with computer program on the computer readable storage medium, which is handled
The step of device executes the training method of above-mentioned Electronic Finance resources model when running.
Specifically, which can be general storage medium, such as mobile disk, hard disk, on the storage medium
Computer program when being run, be able to carry out the training method of above-mentioned Electronic Finance resources model, it is existing to solve
The low problem of technology Electronic Finance resources accuracy, the application is when building model, by history media data and going through
History transaction data is handled, and is increased on the influential feature of Electronic Finance resource tool so that structure model needed becomes certainly
Amount is more diversified, and it is pre- to effectively improve Electronic Finance resource for influence of the comprehensive considering various effects to Electronic Finance resource fluctuations
The accuracy of survey.
Corresponding to the Electronic Finance resource trends prediction technique in Fig. 2, the embodiment of the present application also provides a kind of computers
Equipment 600, as shown in fig. 6, the equipment includes memory 601, processor 602 and is stored on the memory 601 and can be at this
The computer program run on processor 602, wherein above-mentioned processor 602 realizes above-mentioned gold when executing above computer program
Melt e-sourcing trend forecasting method.
Specifically, above-mentioned memory 601 and processor 602 can be general memory and processor, do not do have here
Body limits, and when the computer program of 602 run memory 601 of processor storage, is able to carry out above-mentioned Electronic Finance resource and becomes
Gesture prediction technique, to solve the problems, such as that prior art Electronic Finance resources accuracy is low, the application builds Electronic Finance
Influence of the resources model comprehensive considering various effects to Electronic Finance resource fluctuations, effectively improves Electronic Finance resources
Accuracy.
Corresponding to the Electronic Finance resource trends prediction technique in Fig. 2, the embodiment of the present application also provides a kind of computers
Readable storage medium storing program for executing is stored with computer program on the computer readable storage medium, which is run by processor
The step of Shi Zhihang above-mentioned Electronic Finance resource trends prediction techniques.
Specifically, which can be general storage medium, such as mobile disk, hard disk, on the storage medium
Computer program when being run, above-mentioned Electronic Finance resource trends prediction technique is able to carry out, to solve prior art gold
Melt the low problem of e-sourcing prediction accuracy, the application builds Electronic Finance resources model comprehensive considering various effects pair
The influence of Electronic Finance resource fluctuations effectively improves the accuracy of Electronic Finance resources.
In embodiment provided herein, it should be understood that disclosed system and method, it can be by others side
Formula is realized.System embodiment described above is only schematical, for example, the division of the unit, only one kind are patrolled
Volume function divides, formula that in actual implementation, there may be another division manner, in another example, multiple units or component can combine or can
To be integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Coupling, direct-coupling or communication connection can be INDIRECT COUPLING or communication link by some communication interfaces, system or unit
It connects, can be electrical, machinery or other forms.
The unit illustrated as separating component may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, you can be located at a place, or may be distributed over multiple
In network element.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in embodiment provided by the present application can be integrated in a processing unit, also may be used
It, can also be during two or more units be integrated in one unit to be that each unit physically exists alone.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in a computer read/write memory medium.Based on this understanding, the technical solution of the application is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be expressed in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be
People's computer, server or network equipment etc.) execute each embodiment the method for the application all or part of step.
And storage medium above-mentioned includes:USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic disc or CD.
It should be noted that:Similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined, then it further need not be defined and explained in subsequent attached drawing in a attached drawing, in addition, term " the
One ", " second ", " third " etc. are only used for distinguishing description, are not understood to indicate or imply relative importance.
Finally it should be noted that:Embodiment described above, the only specific implementation mode of the application, to illustrate the application
Technical solution, rather than its limitations, the protection domain of the application is not limited thereto, although with reference to the foregoing embodiments to this Shen
It please be described in detail, it will be understood by those of ordinary skill in the art that:Any one skilled in the art
In the technical scope that the application discloses, it can still modify to the technical solution recorded in previous embodiment or can be light
It is readily conceivable that variation or equivalent replacement of some of the technical features;And these modifications, variation or replacement, do not make
The essence of corresponding technical solution is detached from the spirit and scope of the embodiment of the present application technical solution.The protection in the application should all be covered
Within the scope of.Therefore, the protection domain of the application shall be subject to the protection scope of the claim.
Claims (10)
1. a kind of training method of Electronic Finance resources model, which is characterized in that this method includes:
The history media data of publisher based on Electronic Finance resource determines that the medium property of the Electronic Finance resource is special
Sign;
Based on the historical trading data of the Electronic Finance resource, the transaction attributive character of the Electronic Finance resource is determined;
Screening Treatment is carried out to the medium property feature and the transaction attributive character, obtains influencing Electronic Finance resource fluctuations
Key feature;
Using the key feature as independent variable, the fluctuation characteristic of the Electronic Finance resource fluctuations situation will be characterized as because becoming
Amount builds at least two preset models and is trained;
Based on the prediction result that at least two preset model obtains, using preset model fusion method to described two default
Model carries out fusion treatment, obtains Electronic Finance resources model.
2. the method as described in claim 1, which is characterized in that the prediction knot obtained based at least two preset model
Fruit carries out fusion treatment at least two preset model using preset model fusion method, specifically includes:
Using the prediction result of at least two preset model as independent variable, using the fluctuation characteristic as dependent variable, structure
Fusion Model is simultaneously trained.
3. the method as described in claim 1, which is characterized in that it is described using key feature as independent variable, by fluctuation spy
Sign is used as dependent variable, builds at least two preset models and is trained, specifically includes:
Based on the history media data and historical trading data, the characteristic value of the key feature is determined;
Based on the historical volatility data, the characteristic value of fluctuation characteristic is determined;Wherein, the characteristic value of the fluctuation characteristic is corresponding
Historical time compares the corresponding historical time lag of characteristic value of the key feature;
Determine at least two preset models;And
For each preset model, using the characteristic value of the key feature as the value of independent variable, by corresponding fluctuation characteristic
Value of the characteristic value as dependent variable, is trained the preset model, obtains at least two preset models for completing training.
4. method as claimed in claim 3, which is characterized in that at least two preset model includes:Neural network prediction
Model;
For each preset model, using the characteristic value of key feature as the value of independent variable, by the characteristic value of the fluctuation characteristic
As the value of dependent variable, which is trained, is specifically included:
For multiple neural network models, following training operation is executed respectively, wherein multiple neural networks have different nerves
The network number of plies:
It is right using the characteristic value of the fluctuation characteristic as the value of dependent variable using the characteristic value of key feature as the value of independent variable
Current Situation of Neural Network model is trained, and obtains the index value of the pre-set level for weighing model prediction accuracy;
Using the highest neural network model of index value as finally determining neural network prediction model.
5. the method as described in claim 1, which is characterized in that the described pair of determining medium property feature and the transaction
Attributive character carries out Screening Treatment, obtains the key feature for influencing Electronic Finance resource fluctuations, including:
The medium property feature and the transaction attributive character degree of correlation between the fluctuation characteristic respectively are calculated, by phase
Guan Du is more than that the medium property feature of default relevance threshold and/or transaction attributive character are determined as the first effect characteristics collection;
Screening Treatment is carried out to the medium property feature and the transaction attributive character using decision tree Variable Selection method, is obtained
To the second effect characteristics collection;
The first effect characteristics collection and the second effect characteristics collection are handled using preset characteristic value Processing Algorithm,
Obtain third effect characteristics collection;
The expansion factor value that third effect characteristics concentrate each third effect characteristics is calculated, by expansion factor value less than default expansion
The third effect characteristics of factor threshold are determined as the key feature.
6. method as claimed in claim 5, which is characterized in that described to use preset characteristic value Processing Algorithm to described first
Effect characteristics collection and the second effect characteristics collection are handled, and third effect characteristics collection is obtained, including:
The historical trading number of the history media data of publisher based on the Electronic Finance resource and the Electronic Finance resource
According to, determine first effect characteristics concentrate each first effect characteristics characteristic value and second effect characteristics concentrate each second
The characteristic value of effect characteristics;
Using the characteristic value of each first effect characteristics and the characteristic value of each second effect characteristics as independent variable
Value inputs the characteristic value that training is completed in advance and handles model, obtains the notable of each first effect characteristics and each second effect characteristics
Degree;
Significance is more than corresponding first effect characteristics of significance threshold value or the second effect characteristics are determined as the third and influence
Feature set.
7. a kind of Electronic Finance resource trends prediction technique, which is characterized in that this method includes:
The history media data of publisher based on Electronic Finance resource to be predicted determines the Electronic Finance resource to be predicted
The characteristic value of medium property feature;And
Based on the historical trading data of the Electronic Finance resource to be predicted, the transaction of the Electronic Finance resource to be predicted is determined
The characteristic value of attributive character;
Using the characteristic value of the characteristic value of determining medium property feature and transaction attributive character as the value of independent variable, input is as weighed
Profit requires the Electronic Finance resources model that any one of 1-6 is determined, predicts the rise probability of the Electronic Finance resource.
8. a kind of training device of Electronic Finance resources model, which is characterized in that the device includes:
First determining module is used for the history media data of the publisher based on Electronic Finance resource, determines the Electronic Finance
The medium property feature of resource;
Second determining module is used for the historical trading data based on the Electronic Finance resource, determines the Electronic Finance resource
Transaction attributive character;
Screening module obtains influencing gold for carrying out Screening Treatment to the medium property feature and the transaction attributive character
Melt the key feature of e-sourcing fluctuation;
Module is built, for using the key feature as independent variable, the wave of the Electronic Finance resource fluctuations situation will to be characterized
Dynamic feature builds at least two preset models and is trained as dependent variable;
Processing module, the prediction result for being obtained based at least two preset model, using preset model fusion method
Fusion treatment is carried out to described two preset models, obtains Electronic Finance resources model.
9. a kind of Electronic Finance resource trends prediction meanss, which is characterized in that the device includes:
First determining module is used for the history media data of the publisher based on Electronic Finance resource to be predicted, is waited for described in determination
Predict the characteristic value of the medium property feature of Electronic Finance resource;
Second determining module is used for the historical trading data based on the Electronic Finance resource to be predicted, determines described to be predicted
The characteristic value of the transaction attributive character of Electronic Finance resource
Prediction module, the characteristic value of the medium property feature for will determine and the characteristic value of transaction attributive character are as independent variable
Value, the input Electronic Finance resources model that such as any one of claim 1-6 is determined predicts the Electronic Finance resource
Rise probability.
10. a kind of computer equipment includes memory, processor and is stored on the memory and can be on the processor
The computer program of operation, which is characterized in that the processor realized when executing the computer program the claims 1 to
The step of 6 any one of them method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810398234.4A CN108596393A (en) | 2018-04-28 | 2018-04-28 | A kind of training method and device of Electronic Finance resources model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810398234.4A CN108596393A (en) | 2018-04-28 | 2018-04-28 | A kind of training method and device of Electronic Finance resources model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108596393A true CN108596393A (en) | 2018-09-28 |
Family
ID=63610586
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810398234.4A Pending CN108596393A (en) | 2018-04-28 | 2018-04-28 | A kind of training method and device of Electronic Finance resources model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108596393A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109697532A (en) * | 2018-12-26 | 2019-04-30 | 北京量子保科技有限公司 | A kind of Driving Test insures real-time air control model training method and electronic equipment |
CN110175637A (en) * | 2019-05-09 | 2019-08-27 | 北京工商大学 | Non-stationary time-series data depth prediction technique, system, storage medium and equipment |
CN110334936A (en) * | 2019-06-28 | 2019-10-15 | 阿里巴巴集团控股有限公司 | A kind of construction method, device and the equipment of credit qualification Rating Model |
CN110443374A (en) * | 2019-08-14 | 2019-11-12 | 腾讯科技(深圳)有限公司 | A kind of resource information processing method, device and equipment |
CN111932379A (en) * | 2020-09-23 | 2020-11-13 | 北京口袋财富信息科技有限公司 | Data processing method and device, electronic equipment and readable storage medium |
WO2021073152A1 (en) * | 2019-10-14 | 2021-04-22 | 平安科技(深圳)有限公司 | Data label generation method and apparatus based on neural network, and terminal and medium |
CN113689237A (en) * | 2021-08-20 | 2021-11-23 | 北京达佳互联信息技术有限公司 | Method and device for determining to-be-launched media resource and media resource processing model |
-
2018
- 2018-04-28 CN CN201810398234.4A patent/CN108596393A/en active Pending
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109697532A (en) * | 2018-12-26 | 2019-04-30 | 北京量子保科技有限公司 | A kind of Driving Test insures real-time air control model training method and electronic equipment |
CN110175637A (en) * | 2019-05-09 | 2019-08-27 | 北京工商大学 | Non-stationary time-series data depth prediction technique, system, storage medium and equipment |
CN110334936A (en) * | 2019-06-28 | 2019-10-15 | 阿里巴巴集团控股有限公司 | A kind of construction method, device and the equipment of credit qualification Rating Model |
CN110334936B (en) * | 2019-06-28 | 2023-09-29 | 创新先进技术有限公司 | Method, device and equipment for constructing credit qualification scoring model |
CN110443374A (en) * | 2019-08-14 | 2019-11-12 | 腾讯科技(深圳)有限公司 | A kind of resource information processing method, device and equipment |
WO2021073152A1 (en) * | 2019-10-14 | 2021-04-22 | 平安科技(深圳)有限公司 | Data label generation method and apparatus based on neural network, and terminal and medium |
CN111932379A (en) * | 2020-09-23 | 2020-11-13 | 北京口袋财富信息科技有限公司 | Data processing method and device, electronic equipment and readable storage medium |
CN113689237A (en) * | 2021-08-20 | 2021-11-23 | 北京达佳互联信息技术有限公司 | Method and device for determining to-be-launched media resource and media resource processing model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108596393A (en) | A kind of training method and device of Electronic Finance resources model | |
Mahdiraji et al. | Marketing strategies evaluation based on big data analysis: a CLUSTERING-MCDM approach | |
US8166155B1 (en) | System and method for website experimentation | |
CN108256691A (en) | Refund Probabilistic Prediction Model construction method and device | |
CN107220217A (en) | Characteristic coefficient training method and device that logic-based is returned | |
CN108182634A (en) | A kind of training method for borrowing or lending money prediction model, debt-credit Forecasting Methodology and device | |
CN107066616A (en) | Method, device and electronic equipment for account processing | |
CN104657412A (en) | Recommendation system for specifying and achieving goals | |
CN107798607A (en) | Asset Allocation strategy acquisition methods, device, computer equipment and storage medium | |
CN109766454A (en) | A kind of investor's classification method, device, equipment and medium | |
CN109345050A (en) | A kind of quantization transaction prediction technique, device and equipment | |
Liu et al. | Novel evolutionary multi-objective soft subspace clustering algorithm for credit risk assessment | |
CN107633455A (en) | Credit estimation method and device based on data model | |
CN110135976A (en) | User's portrait generation method, device, electronic equipment and computer-readable medium | |
CN108734587A (en) | The recommendation method and terminal device of financial product | |
Huang et al. | Improve reputation evaluation of crowdsourcing participants using multidimensional index and machine learning techniques | |
CN108596765A (en) | A kind of Electronic Finance resource recommendation method and device | |
JPWO2018142753A1 (en) | Information processing apparatus, information processing method, and information processing program using deep learning | |
CN110222733A (en) | The high-precision multistage neural-network classification method of one kind and system | |
CN107346515A (en) | A kind of credit card Forecasting Methodology and device by stages | |
Moslehi et al. | Analyzing and investigating the use of electronic payment tools in Iran using data mining techniques | |
Mukherjee et al. | Detection of defaulters in P2P lending platforms using unsupervised learning | |
CN112132589A (en) | Method for constructing fraud recognition model based on multiple times of fusion | |
Dostál et al. | Business applications of fuzzy logic | |
Dash et al. | A legendre neural network for credit card fraud detection |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180928 |