CN109948905A - A kind of payment classification of risks method and device based on subscriber payment data - Google Patents
A kind of payment classification of risks method and device based on subscriber payment data Download PDFInfo
- Publication number
- CN109948905A CN109948905A CN201910131790.XA CN201910131790A CN109948905A CN 109948905 A CN109948905 A CN 109948905A CN 201910131790 A CN201910131790 A CN 201910131790A CN 109948905 A CN109948905 A CN 109948905A
- Authority
- CN
- China
- Prior art keywords
- payment
- user
- risk
- data
- history
- 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
Abstract
The payment classification of risks method and device based on subscriber payment data that this application discloses a kind of obtains the history payment data of user;It is paid the fees data according to history, marks the payment risk classifications of user, payment risk classifications include payment risk and without payment risk;According to the history of tape label payment data, logistic regression algorithm is used to construct payment two disaggregated model of risk;According to payment two disaggregated model of risk, the payment risk classifications for generating the user of new payment data are judged.The technical solution of the application is realized according to two disaggregated models to the payment classification of risks of user, is improved the accuracy and reliability of electricity charge risk management, is reduced the sale of electricity risk of grid company.
Description
Technical field
This application involves analysis and survey control technology field more particularly to a kind of payment wind based on subscriber payment data
Dangerous classification method and device.
Background technique
With the continuous development of electric power enterprise, the group of power customer also increasingly grows.But the credit of client is presented
Irregular situation, so that electric power enterprise encounters serious electrically charge problem, the especially meeting of power customer Credit Deficiency
Cause the electricity charge that can not pay neat situation, not only brings difficulty to the operation of electric power enterprise, also seriously destroying electricity market just
Normal transaction order increases the business risk of electric power enterprise.
In order to reduce sale of electricity risk, whether owed in the past currently, grid company relies primarily on staff according to power customer
The data such as expense and arrearage number carry out the judgement of payment risk, and not only time-consuming and laborious efficiency is inefficient, but also judgment method is more
It is extensive and subjective, so that judging result is not accurate enough and reliable.Therefore, work about electric power personnel how are avoided to pay power customer
Take the phenomenon that risk judgment is more extensive and subjective, result is inaccurate, becomes those skilled in the art's urgent problem to be solved.
Summary of the invention
The payment classification of risks method and device based on subscriber payment data that this application provides a kind of, avoids work about electric power
Payment risk judgment more extensive and subjective, result inaccurate phenomenon of the personnel to power customer.
On the one hand, the embodiment of the present application provides a kind of payment classification of risks method based on subscriber payment data, comprising:
The history for obtaining user is paid the fees data, history payment data include subscriber payment number, subscriber arrearage number,
User's cash payment number, user self-help payment number, user's collection payment number, user add up arrearage, user's electricity, user
The average number of days of payment and average electricity price;
According to history payment data, the payment risk classifications of the user are marked, the payment risk classifications include
There is payment risk and without payment risk;
According to the history of tape label payment data, logistic regression algorithm is used to construct payment two disaggregated model of risk;
According to two disaggregated model of payment risk, the payment risk classifications for generating the user of new payment data are judged.
With reference to first aspect, described according to the history of tape label payment data, use logistic regression algorithm to construct payment wind
Nearly the step of two disaggregated model includes:
Using the method for random sampling, the history payment data of the tape label are divided into training data and confirmation data;
According to the training data, calculates and assume function H (X);
According to the training data and assume function, calculate cost function cost (W), finds when cost function minimum pair
The weight vectors answered;
The weight vectors are optimized using gradient descent algorithm;
Using the weight vectors after optimization, the hypothesis function H (X) is updated.
With reference to first aspect, the method also includes:
Classified using two disaggregated model of payment risk to the confirmation data;
Obtain the classification results of the corresponding confirmation data of difference candidate parameter in candidate learning rate array, the candidate study
It include several candidate study rate scores in rate array;
According to the classification results, the final learning rate for determining and using is selected.
With reference to first aspect, described according to payment two disaggregated model of risk, judge the user's for generating new payment data
Pay the fees risk classifications the step of include:
Weight vectors after optimizing described in the corresponding characteristic use of each data in new payment data are weighted
Summation;
According to the weighted sum value of acquisition, the hypothesis function is calculated, obtains the payment value-at-risk of user;
According to the payment value-at-risk, the payment risk classifications for generating the user of target-seeking payment data are judged.
With reference to first aspect, described according to the payment value-at-risk, judge the payment for generating the user of target-seeking payment data
The step of risk classifications includes:
If the payment value-at-risk is greater than 0.5, the user is the user for having payment risk;
If the payment value-at-risk, less than 0.5, the user is the user without payment risk.
With reference to first aspect, the method also includes: normalizing is carried out to each data in the history of user payment data
Change processing, each data are mapped between [0,1].
With reference to first aspect, the hypothesis function H (X) is calculated according to the following formula:
Wherein, X indicates the feature vector (x of data in user's history payment data0,x1,x2,x3,x4,x5,x6,x7,x8,
x9), W indicates weight vectors (w corresponding with described eigenvector0, w1,w2,w3,w4,w5,w6,w7,w8,w9) and feature x0
Perseverance is 1, feature x1、x2、x3、x4、x5、x6、x7、x8And x9It is corresponding in turn to subscriber payment number, subscriber arrearage number, user's cash
Payment number, user self-help payment number, user's collection payment number, the accumulative arrearage of user, user's electricity, subscriber payment are average
Number of days and average electricity price, w0、w1、w2、w3、w4、w5、w6、w7、w8And w9It is corresponding in turn to feature x1、x2、x3、x4、x5、x6、x7、x8With
x9。
With reference to first aspect, the cost function cost (W) is calculated according to the following formula:
Wherein, W indicates that weight vectors corresponding with feature vector, m indicate the quantity of user, and y indicates the payment of user
Risk classifications, have the corresponding y of payment risk labeled as 1, and the corresponding y of no payment risk is labeled as 0.
With reference to first aspect, described to be carried out using the gradient descent algorithm optimization weight vectors according to following formula:
On the other hand, the embodiment of the present application provides a kind of payment classification of risks device based on subscriber payment data, packet
It includes:
History payment data capture unit, for obtaining the history payment data of user, the history payment data include
Subscriber payment number, subscriber arrearage number, user's cash payment number, user self-help payment number, user's collection payment number,
User adds up arrearage, user's electricity, subscriber payment are averaged number of days and average electricity price;
Risk classifications marking unit, for marking the payment risk classifications of the user according to history payment data,
The payment risk classifications include payment risk and without payment risk;
Two disaggregated model construction units, for being constructed using logistic regression algorithm according to the history of tape label payment data
Payment two disaggregated model of risk;
Payment classification of risks unit, for judging to generate new payment data according to two disaggregated model of payment risk
User payment risk classifications.
From the above technical scheme, the embodiment of the present application provides a kind of payment risk based on subscriber payment data point
Class method and device obtains the history payment data of user;According to history payment data, the payment risk classifications of user are marked,
Payment risk classifications include payment risk and without payment risk;According to the history of tape label payment data, logistic regression is used
Algorithm building payment two disaggregated model of risk;According to payment two disaggregated model of risk, the user for generating new payment data is judged
Payment risk classifications.The technical solution of the application is realized according to two disaggregated models to the payment classification of risks of user, is improved
The accuracy and reliability of electricity charge risk management reduces the sale of electricity risk of grid company.
Detailed description of the invention
In order to illustrate more clearly of the technical solution of the application, attached drawing needed in case study on implementation will be made below
Simply introduce, it should be apparent that, for those of ordinary skills, in the premise of not making the creative labor property
Under, it is also possible to obtain other drawings based on these drawings.
Fig. 1 is a kind of process of payment classification of risks method based on subscriber payment data provided by the embodiments of the present application
Figure;
Fig. 2 is a kind of structural frames of payment classification of risks device based on subscriber payment data provided by the embodiments of the present application
Figure.
Specific embodiment
In order to make those skilled in the art better understand the technical solutions in the application, below in conjunction with attached drawing, it is right
Technical solution in the embodiment of the present application is clearly and completely described.
Referring to Fig. 1, the embodiment of the present application provides a kind of payment classification of risks method based on subscriber payment data, packet
It includes:
Step 101, the history payment data of user are obtained, the history payment data include subscriber payment number, user
Arrearage number, user's cash payment number, user self-help payment number, user's collection payment number, user add up arrearage, user
Electricity, subscriber payment are averaged number of days and average electricity price;Further, user's cash payment, user self-help payment and user's collection
Payment refers to three kinds of channels of user's electricity payment, and the subscriber payment number of days that is averaged refers to that paying the electricity charge from notice pays electricity to user is practical
Average time between taking.
Step 102, according to history payment data, the payment risk classifications of the user, the payment risk are marked
Type include payment risk and without payment risk.
Step 103, according to the history of tape label payment data, logistic regression algorithm is used to construct the payment classification mould of risk two
Type.
Optionally, described according to the history of tape label payment data, payment risk two, which is constructed, using logistic regression algorithm divides
The step of class model includes:
Step 201, using the method for random sampling, the history payment data of the tape label are divided into training data and really
Recognize data.Training data is used to select the suitable learning rate of two disaggregated models, leads to for constructing two disaggregated models, confirmation data
Often the ratio of the training data after sampling and confirmation data is 8:2.
Step 202, it according to the training data, calculates and assumes function H (X);Further, hypothesis function H (X) basis
Following formula is calculated:
Wherein, X indicates the feature vector (x of data in user's history payment data0,x1,x2,x3,x4,x5,x6,x7,x8,
x9), W indicates weight vectors (w corresponding with described eigenvector0,w1,w2,w3,w4,w5,w6,w7,w8,w9) and feature x0
Perseverance is 1, feature x1、x2、x3、x4、x5、x6、x7、x8And x9It is corresponding in turn to subscriber payment number, subscriber arrearage number, user's cash
Payment number, user self-help payment number, user's collection payment number, the accumulative arrearage of user, user's electricity, subscriber payment are average
Number of days and average electricity price, w0、w1、w2、w3、w4、w5、w6、w7、w8And w9It is corresponding in turn to feature x1、x2、x3、x4、x5、x6、x7、x8With
x9。
Step 203, it according to the training data and hypothesis function, calculates cost function cost (W), finds and work as cost function
Corresponding weight vectors when minimum;Further, the cost function cost (W) is calculated according to the following formula:
Wherein, W indicates that weight vectors corresponding with feature vector, m indicate the quantity of user, and y indicates the payment of user
Risk classifications, have the corresponding y of payment risk labeled as 1, and the corresponding y of no payment risk is labeled as 0.
Step 204, the weight vectors are optimized using gradient descent algorithm;Further, described to use gradient descent algorithm
Optimize the weight vectors to carry out according to following formula:
Gradient descent algorithm be actually using iteration by the way of update W, wherein the number of iterations is at most set as
10000 times, cost function cost (W) minimum value is set as 0.01.
Step 205, using the weight vectors after optimization, the hypothesis function H (X) is updated.
Step 104, according to two disaggregated model of payment risk, judge the payment wind for generating the user of new payment data
Dangerous type.
Optionally, described according to payment two disaggregated model of risk, judge the payment wind for generating the user of new payment data
The step of dangerous type includes:
Step 301, the weight vectors after optimizing described in the corresponding characteristic use of each data in new payment data
It is weighted summation;The corresponding characteristic weighing of payment data that will be new is summed, and h1=w is obtained0+x1w1+x2w2+x3w3+x4w4+
x5w5+x6w6+x7w7+x8w8+x9w9。
Step 302, according to the weighted sum value of acquisition, the hypothesis function is calculated, obtains the payment value-at-risk of user;This
When the hypothesis function that calculates be H (h1), the value of the hypothesis function is also payment value-at-risk.
Step 303, according to the payment value-at-risk, judge the payment risk classifications for generating the user of target-seeking payment data.
Optionally, described according to the payment value-at-risk, judge the payment risk class for generating the user of target-seeking payment data
The step of type includes:
Step 401, if the payment value-at-risk H (h1) is greater than 0.5, the user is the user for having payment risk;
Step 402, if the payment value-at-risk H (h1) is less than 0.5, the user is the user without payment risk.
In addition, if the value is in the gray zone of judgement when the payment value-at-risk H (h1) is equal to 0.5, it can not be straight
Connect the risk classifications for judging user, it is also necessary to carry out other processing.
Optionally, the method also includes:
Step 501, classified using two disaggregated model of payment risk to the confirmation data.
Step 502, the classification results of the corresponding confirmation data of difference candidate parameter in candidate learning rate array are obtained, it is described
It include several candidate study rate scores in candidate learning rate array.
Step 503, according to the classification results, the final learning rate for determining and using is selected.
Different candidate parameters in confirmation data comparison candidate's learning rate data (0.1,0.01,0.001) used herein
Classifying quality in confirmation data, selects the corresponding parameter of best classifying quality.It finally obtains, when learning rate is set as 0.1
When, classifying quality is optimal.
Optionally, the method also includes: each data in the history of user payment data are normalized,
Each data are mapped between [0,1].
Specifically, the formula used during normalization are as follows:Wherein, the value difference of i
For 1-9.
Referring to fig. 2, the embodiment of the present application also provides a kind of payment classification of risks device based on subscriber payment data, packet
It includes:
History payment data capture unit 21, for obtaining the history payment data of user, the history payment data packet
Include subscriber payment number, subscriber arrearage number, user's cash payment number, user self-help payment number, user's collection payment time
Number, user add up arrearage, user's electricity, subscriber payment are averaged number of days and average electricity price;
Risk classifications marking unit 22, for marking the payment risk class of the user according to history payment data
Type, the payment risk classifications include payment risk and without payment risk;
Two disaggregated model construction units 23, for using logistic regression algorithm structure according to the history of tape label payment data
Build payment two disaggregated model of risk;
Payment classification of risks unit 24, for judging to generate new payment number according to two disaggregated model of payment risk
According to user payment risk classifications.
Optionally, the two disaggregated models construction unit 23 includes:
The history payment data of the tape label are divided into trained number for the method using random sampling by sampling unit
According to confirmation data;
Assuming that function calculating unit, for calculating and assuming function H (X) according to the training data;
Cost function calculation unit, for calculating cost function cost (W) according to the training data and hypothesis function,
Find the corresponding weight vectors when cost function minimum;
Optimize unit, for optimizing the weight vectors using gradient descent algorithm;
Updating unit updates the hypothesis function H (X) for using the weight vectors after optimizing.
Optionally, described device further includes learning rate determination unit, is used for:
Classified using two disaggregated model of payment risk to the confirmation data;
Obtain the classification results of the corresponding confirmation data of difference candidate parameter in candidate learning rate array, the candidate study
It include several candidate study rate scores in rate array;
According to the classification results, the final learning rate for determining and using is selected.
Optionally, the payment classification of risks unit 24 includes:
Summation unit, for the weight after optimizing described in the corresponding characteristic use of each data in new payment data
Vector is weighted summation;
Value-at-risk acquiring unit calculates the hypothesis function, obtains paying for user for the weighted sum value according to acquisition
Take value-at-risk;
Judging unit, the payment risk of the user for according to the payment value-at-risk, judging to generate target-seeking payment data
Type.
Optionally, the judging unit is also used to:
If the payment value-at-risk is greater than 0.5, the user is the user for having payment risk;
If the payment value-at-risk, less than 0.5, the user is the user without payment risk.
Optionally, described device further includes normalization unit, is used for: to each data in the history payment data of user
It is normalized, each data is mapped between [0,1].
Optionally, the hypothesis function calculating unit calculates according to the following formula assumes function H (X):
Wherein, X indicates the feature vector (x of data in user's history payment data0,x1,x2,x3,x4,x5,x6,x7,x8,
x9), W indicates weight vectors (w corresponding with described eigenvector0, w1,w2,w3,w4,w5,w6,w7,w8,w9) and feature x0
Perseverance is 1, feature x1、x2、x3、x4、x5、x6、x7、x8And x9It is corresponding in turn to subscriber payment number, subscriber arrearage number, user's cash
Payment number, user self-help payment number, user's collection payment number, the accumulative arrearage of user, user's electricity, subscriber payment are average
Number of days and average electricity price, w0、w1、w2、w3、w4、w5、w6、w7、w8And w9It is corresponding in turn to feature x1、x2、x3、x4、x5、x6、x7、x8With
x9。
Optionally, the cost function calculation unit calculates cost function cost (W) according to the following formula:
Wherein, W indicates that weight vectors corresponding with feature vector, y indicate the payment risk classifications of user, there is payment wind
The corresponding y in danger is labeled as 1, and the corresponding y of no payment risk is labeled as 0.
Optionally, the optimization unit optimizes weight vectors according to following formula:
From the above technical scheme, the embodiment of the present application provides a kind of payment risk based on subscriber payment data point
Class method and device obtains the history payment data of user;According to history payment data, the payment risk classifications of user are marked,
Payment risk classifications include payment risk and without payment risk;According to the history of tape label payment data, logistic regression is used
Algorithm building payment two disaggregated model of risk;According to payment two disaggregated model of risk, the user for generating new payment data is judged
Payment risk classifications.The technical solution of the application is realized according to two disaggregated models to the payment classification of risks of user, is improved
The accuracy and reliability of electricity charge risk management reduces the sale of electricity risk of grid company.
The application can be used in numerous general or special purpose computing system environments or configuration.Such as: personal computer, service
Device computer, handheld device or portable device, laptop device, multicomputer system, microprocessor-based system, top set
Box, programmable consumer-elcetronics devices, network PC, minicomputer, mainframe computer, including any of the above system or equipment
Distributed computing environment etc..
The application can describe in the general context of computer-executable instructions executed by a computer, such as program
Module.Generally, program module includes routines performing specific tasks or implementing specific abstract data types, programs, objects, group
Part, data structure etc..The application can also be practiced in a distributed computing environment, in these distributed computing environments, by
Task is executed by the connected remote processing devices of communication network.In a distributed computing environment, program module can be with
In the local and remote computer storage media including storage equipment.
Those skilled in the art will readily occur to its of the application after considering specification and practicing application disclosed herein
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the application, these modifications, purposes or
Person's adaptive change follows the general principle of the application and including the undocumented common knowledge in the art of the application
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the application are by following
Claim is pointed out.
It should be understood that the application is not limited to the precise structure that has been described above and shown in the drawings, and
And various modifications and changes may be made without departing from the scope thereof.Scope of the present application is only limited by the accompanying claims.
Claims (10)
1. a kind of payment classification of risks method based on subscriber payment data characterized by comprising
The history payment data of user are obtained, the history payment data include subscriber payment number, subscriber arrearage number, user
Cash payment number, user self-help payment number, user's collection payment number, user add up arrearage, user's electricity, subscriber payment
Average number of days and average electricity price;
According to history payment data, the payment risk classifications of the user are marked, the payment risk classifications include to pay
Take risk and without payment risk;
According to the history of tape label payment data, logistic regression algorithm is used to construct payment two disaggregated model of risk;
According to two disaggregated model of payment risk, the payment risk classifications for generating the user of new payment data are judged.
2. payment classification of risks method according to claim 1, which is characterized in that described to be paid the fees according to the history of tape label
Data, constructing the step of paying the fees two disaggregated model of risk using logistic regression algorithm includes:
Using the method for random sampling, the history payment data of the tape label are divided into training data and confirmation data;
According to the training data, calculates and assume function H (X);
According to the training data and assume function, calculates cost function cost (W), find corresponding when cost function minimum
Weight vectors;
The weight vectors are optimized using gradient descent algorithm;
Using the weight vectors after optimization, the hypothesis function H (X) is updated.
3. payment classification of risks method according to claim 2, which is characterized in that the method also includes:
Classified using two disaggregated model of payment risk to the confirmation data;
Obtain the classification results of the corresponding confirmation data of difference candidate parameter in candidate learning rate array, candidate's learning rate number
It include several candidate study rate scores in group;
According to the classification results, the final learning rate for determining and using is selected.
4. payment classification of risks method according to claim 2, which is characterized in that described according to the payment classification mould of risk two
Type judges that the step of generating the payment risk classifications of the user of new payment data includes:
Weight vectors after optimizing described in the corresponding characteristic use of each data in new payment data are weighted summation;
According to the weighted sum value of acquisition, the hypothesis function is calculated, obtains the payment value-at-risk of user;
According to the payment value-at-risk, the payment risk classifications for generating the user of target-seeking payment data are judged.
5. payment classification of risks method according to claim 4, which is characterized in that it is described according to the payment value-at-risk,
Judge that the step of generating the payment risk classifications of the user of target-seeking payment data includes:
If the payment value-at-risk is greater than 0.5, the user is the user for having payment risk;
If the payment value-at-risk, less than 0.5, the user is the user without payment risk.
6. payment classification of risks method described in -5 according to claim 1, which is characterized in that the method also includes:
Each data in the history payment data of user are normalized, each data are mapped between [0,1].
7. payment classification of risks method according to claim 2, which is characterized in that the hypothesis function H (X) is according to following
Formula is calculated:
Wherein, X indicates the feature vector (x of data in user's history payment data0,x1,x2,x3,x4,x5,x6,x7,x8,x9), W table
Show weight vectors (w corresponding with described eigenvector0,w1,w2,w3,w4,w5,w6,w7,w8,w9) and feature x0Perseverance is 1,
Feature x1、x2、x3、x4、x5、x6、x7、x8And x9It is corresponding in turn to subscriber payment number, subscriber arrearage number, user's cash payment
Number, user self-help payment number, user's collection payment number, user add up arrearage, user's electricity, subscriber payment are averaged number of days and
Average electricity price, w0、w1、w2、w3、w4、w5、w6、w7、w8And w9It is corresponding in turn to feature x1、x2、x3、x4、x5、x6、x7、x8And x9。
8. payment classification of risks method according to claim 7, which is characterized in that cost function cost (W) basis
Following formula is calculated:
Wherein, W indicates that weight vectors corresponding with feature vector, m indicate the quantity of user, and y indicates the payment risk of user
Type, has the corresponding y of payment risk labeled as 1, and the corresponding y of no payment risk is labeled as 0.
9. payment classification of risks method according to claim 8, which is characterized in that described to be optimized using gradient descent algorithm
The weight vectors are carried out according to following formula:
10. a kind of payment classification of risks device based on subscriber payment data characterized by comprising
History payment data capture unit, for obtaining the history payment data of user, the history payment data include user
Payment number, subscriber arrearage number, user's cash payment number, user self-help payment number, user's collection payment number, user
Accumulative arrearage, user's electricity, subscriber payment are averaged number of days and average electricity price;
Risk classifications marking unit, it is described for marking the payment risk classifications of the user according to history payment data
Payment risk classifications include payment risk and without payment risk;
Two disaggregated model construction units, for using logistic regression algorithm to construct and paying the fees according to the history of tape label payment data
Two disaggregated model of risk;
Payment classification of risks unit, for judging the use of the new payment data of generation according to two disaggregated model of payment risk
The payment risk classifications at family.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910131790.XA CN109948905A (en) | 2019-02-22 | 2019-02-22 | A kind of payment classification of risks method and device based on subscriber payment data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910131790.XA CN109948905A (en) | 2019-02-22 | 2019-02-22 | A kind of payment classification of risks method and device based on subscriber payment data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109948905A true CN109948905A (en) | 2019-06-28 |
Family
ID=67006977
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910131790.XA Pending CN109948905A (en) | 2019-02-22 | 2019-02-22 | A kind of payment classification of risks method and device based on subscriber payment data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109948905A (en) |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107895245A (en) * | 2017-12-26 | 2018-04-10 | 国网宁夏电力有限公司银川供电公司 | A kind of tariff recovery methods of risk assessment based on user's portrait |
-
2019
- 2019-02-22 CN CN201910131790.XA patent/CN109948905A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107895245A (en) * | 2017-12-26 | 2018-04-10 | 国网宁夏电力有限公司银川供电公司 | A kind of tariff recovery methods of risk assessment based on user's portrait |
Non-Patent Citations (4)
Title |
---|
吴漾等: "基于特征选择改进LR-Bagging算法的电力欠费风险居民客户预测", 《电子产品世界》 * |
周晖等: "基于Logistic回归模型的电力客户欠费违约概率的预测", 《电网技术》 * |
涂莹等: "基于市场细分的逻辑回归模型在电费回收风险预测中的应用研究", 《电力需求侧管理》 * |
邹晓辉: "基于Logistic回归的数据分类问题研究", 《智能计算机与应用》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109919684A (en) | For generating method, electronic equipment and the computer readable storage medium of information prediction model | |
CN106503873A (en) | A kind of prediction user follows treaty method, device and the computing device of probability | |
CN105975483A (en) | User preference-based message pushing method and platform | |
CN110197315A (en) | Methods of risk assessment, device and its storage medium | |
CN108550090A (en) | A kind of processing method and system of determining source of houses pricing information | |
CN107465741A (en) | Information-pushing method and device | |
CN108280757A (en) | User credit appraisal procedure and device | |
CN110084627A (en) | The method and apparatus for predicting target variable | |
CN108256757A (en) | A kind of source of houses conclusion of the business predictor method based on xgboost and estimate platform | |
CN110046981A (en) | A kind of credit estimation method, device and storage medium | |
CN110415106A (en) | A kind of performance analysis forecast assessment system of management body | |
CN107358477A (en) | Tour site hotel real time pricing method, system, storage medium and electronic equipment | |
CN109034490A (en) | A kind of Methods of electric load forecasting, device, equipment and storage medium | |
CN108734567A (en) | A kind of asset management system and its appraisal procedure based on big data artificial intelligence air control | |
CN109993538A (en) | Identity theft detection method based on probability graph model | |
CN109657846A (en) | Power grid alternative subsidy scale impact factor screening technique | |
CN109961328A (en) | The method and apparatus for determining order cooling off period | |
CN110019774A (en) | Label distribution method, device, storage medium and electronic device | |
CN109118282A (en) | A kind of bimodulus mutual inductance intelligent space user draws a portrait management method and terminal | |
CN107222313B (en) | A kind of E-Government cloud service expense calculates monitoring method and system | |
CN113361959A (en) | Method and device for calculating maturity of centralized operation of banking business | |
CN113159874A (en) | Method and device for detecting value-added tax invoice and readable storage medium | |
CN108921425A (en) | A kind of method, system and the server of asset item classifcation of investment | |
CN109948773A (en) | The method and apparatus for generating information | |
CN109948905A (en) | A kind of payment classification of risks method and device based on subscriber payment data |
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 |
Application publication date: 20190628 |
|
RJ01 | Rejection of invention patent application after publication |