CN105279691A - Financial transaction detection method and equipment based on random forest model - Google Patents
Financial transaction detection method and equipment based on random forest model Download PDFInfo
- Publication number
- CN105279691A CN105279691A CN201410361193.3A CN201410361193A CN105279691A CN 105279691 A CN105279691 A CN 105279691A CN 201410361193 A CN201410361193 A CN 201410361193A CN 105279691 A CN105279691 A CN 105279691A
- Authority
- CN
- China
- Prior art keywords
- decision
- variable
- transaction
- random forest
- model
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a financial transaction detection method based on a random forest model. The method includes: (a) obtaining a historical transaction table and a fraudulent transaction table; (b) utilizing the historical transaction table and the fraudulent transaction table to construct a sample data set which includes sample characteristic variables; (c) randomly extracting a plurality of samples from the sample data set with putbacks; and (d) randomly selecting an identical number of characteristic variables for each of the plurality of samples, so as to generate a decision tree model corresponding to the sample and to further generate a random forest model. The invention also discloses financial transaction detection equipment based on the random forest model.
Description
Technical field
The present invention relates to financial transaction fraud detection field, particularly a kind of financial transaction detection method based on Random Forest model and equipment.
Background technology
In traditional bank card fraudulent trading method for detecting, decision-tree model has calculated amount classifying rules that is relatively little, that generate the advantage such as can to understand, and can meet the demand of fraud detection work to a certain extent.But the easy over-fitting of single decision-tree model, classifying rules easily becomes complex, and classification results is unstable.Meanwhile, for the training of unbalanced data, the classification results of decision-tree model is obviously partial to most class, easily causes the inaccurate of classification results.
Summary of the invention
For solving the problem, according to an aspect of the present invention, a kind of financial transaction detection method based on Random Forest model is provided.The method comprises: (a) obtains historical trading table and fraudulent trading table; B () utilizes described historical trading table and described fraudulent trading table to construct sample data sets, described sample data sets comprises sample characteristics variable; C () extracts many increments originally with putting back at random from described sample data sets; D () is the characteristic variable of each this Stochastic choice of the increment equal number in described many increment bases, to generate and this corresponding decision-tree model of this increment, and then generate Random Forest model; E () trains and assesses each of the multiple decision-tree models in described Random Forest model, to obtain the accuracy rate of each decision-tree model; F () determines the ballot weight q of each decision-tree model based on described accuracy rate
i; And (g) utilizes the multiple decision-tree models in described Random Forest model to export y to the response of inputted data of financial transaction
iand described ballot weight q
i, obtain voting results RF according to following formula and judge whether described financial transaction exists swindle:
Wherein, l is the quantity of decision-tree model.
By there being the random sampling of putting back to generate training sample, each sample Stochastic choice equal number characteristic variable participates in training, generates a decision-tree model, decides the classification of concluding the business finally by multiple decision-tree model ballot.This detection method and equipment overcome the shortcoming of single decision-tree model classifying rules complexity, the easy over-fitting of model, classification accuracy instability, this detection method and equipment are for the unbalancedness of bank card business dealing data simultaneously, also have good adaptive faculty.
Said method also can comprise: described transaction, when judging that described financial transaction exists swindle, adds in fraud detection result set by (h); I () confirms the transaction in described fraud detection result set, and the transaction confirming as swindle added in described fraudulent trading table; And (j) re-executes step (a) and (b).
In the above-mentioned methods, step (e) comprises further: the Gini coefficient that in (e1) calculation training sample, all variablees divide in all values; (e2) get Gini coefficient minimum be divided into the first best splitting point; And described training sample divides based on described first best splitting point by (e3), and (e1) and (e2) is repeated respectively to determine the second best splitting point to the training sample after dividing.
In the above-mentioned methods, described sample characteristics variable comprises original variable, context variable and statistical variable.
In the above-mentioned methods, described original variable includes but not limited to, the dealing money directly obtained from described fraudulent trading table and described historical trading table and exchange hour.
In the above-mentioned methods, described context variable includes but not limited to, transaction whether in areal and transaction whether in same trade company.
In the above-mentioned methods, described statistical variable includes but not limited to, certain card number or this card number transaction trade company statistical information within a period of time.
In the above-mentioned methods, the quantity of selected characteristic variable is n, and the relation of the total N of n and characteristic variable is as follows:
In the above-mentioned methods, when the accuracy rate of certain decision-tree model is lower than a threshold value, from described Random Forest model, this decision-tree model is given up.By the assessment to single decision tree, eliminate the decision tree that nicety of grading is too low, give different ballot weights to each decision-tree model simultaneously, thus improve the accuracy of Random Forest model.
According to another aspect of the present invention, provide a kind of financial transaction checkout equipment based on Random Forest model, comprising: for obtaining the device of historical trading table and fraudulent trading table; For utilizing described historical trading table and described fraudulent trading table to construct the device of sample data sets, described sample data sets comprises sample characteristics variable; Many increments device is originally extracted with putting back to for having at random from described sample data sets; For the characteristic variable for each this Stochastic choice of the increment equal number in described many increment bases, to generate and this corresponding decision-tree model of this increment, and then generate the device of Random Forest model; For training and assess each of the multiple decision-tree models in described Random Forest model, to obtain the device of the accuracy rate of each decision-tree model; For determining the ballot weight q of each decision-tree model based on described accuracy rate
idevice; And for utilizing the multiple decision-tree models in described Random Forest model to export y to the response of inputted data of financial transaction
iand described ballot weight q
i, obtain voting results RF according to following formula and judge whether described financial transaction exists the device of swindle:
Wherein, l is the quantity of decision-tree model.
The said equipment also can comprise: for when judging that described financial transaction exists swindle, described transaction is added the device in fraud detection result set; For confirming the transaction in described fraud detection result set, and the transaction confirming as swindle is added the device in described fraudulent trading table.
In the said equipment, for training each of the multiple decision-tree models in described Random Forest model and assess, so that the device obtaining the accuracy rate of each decision-tree model is configured to perform following steps: the Gini coefficient that in (e1) calculation training sample, all variablees divide in all values; (e2) get Gini coefficient minimum be divided into the first best splitting point; And described training sample divides based on described first best splitting point by (e3), and (e1) and (e2) is repeated respectively to determine the second best splitting point to the training sample after dividing.
In the said equipment, described sample characteristics variable comprises original variable, context variable and statistical variable.
In the said equipment, described original variable includes but not limited to, the dealing money directly obtained from described fraudulent trading table and described historical trading table and exchange hour.
In the said equipment, described context variable includes but not limited to, transaction whether in areal and transaction whether in same trade company.
In the said equipment, described statistical variable includes but not limited to, certain card number or this card number transaction trade company statistical information within a period of time.
In the said equipment, the quantity of selected characteristic variable is n, and the relation of the total N of n and characteristic variable is as follows:
The said equipment also can comprise: for when the accuracy rate of certain decision-tree model is lower than a threshold value, give up the device of this decision-tree model from described Random Forest model.
Accompanying drawing explanation
After having read the specific embodiment of the present invention with reference to accompanying drawing, those skilled in the art will become apparent various aspects of the present invention.Those skilled in the art should be understood that: these accompanying drawings only for coordinating embodiment that technical scheme of the present invention is described, and and are not intended to be construed as limiting protection scope of the present invention.
Fig. 1 and Fig. 2 is the embodiment according to the application, based on the schematic flow sheet of the financial transaction detection method of Random Forest model.
Embodiment
Introduce below be of the present invention multiple may some in embodiment, aim to provide basic understanding of the present invention, be not intended to confirm key of the present invention or conclusive key element or limit claimed scope.Easy understand, according to technical scheme of the present invention, do not changing under connotation of the present invention, one of ordinary skill in the art can propose other implementation that can mutually replace.Therefore, following embodiment and accompanying drawing are only the exemplary illustrations to technical scheme of the present invention, and should not be considered as of the present invention all or the restriction be considered as technical solution of the present invention or restriction.
In general, this application provides a kind of financial transaction detection method based on random forest and equipment.By there being the random sampling of putting back to generate training sample, each sample Stochastic choice equal number characteristic variable participates in training, generates a decision-tree model, decides the classification of concluding the business finally by multiple decision-tree model ballot.
The transaction detection method flow scheme design based on random forest of the application as shown in Figure 1.This transaction detection method comprises sampling of data, extraction feature, data prediction, generation sample data, training pattern generation random forest and produces the steps such as transaction.Be specifically described for each step below:
1) sampling of data
Extract all swindle data in fraudulent trading table, from historical trading table, press card number extract transaction data, transaction record is labeled as fraud respectively, normal.Owing to comprising fraudulent trading data in historical trading table, need in the transaction extracted in historical trading table, the card number be included in fraudulent trading table to be rejected.Because fraudulent trading only accounts for a little part in production transaction, when constructing sample data, fraudulent trading and the desirable empirical value 200 (needing with reference to producing actual ratio) of arm's length transaction ratio.
2) feature is extracted
Sample characteristics variable is divided into original variable, context variable, statistical variable.Original variable directly obtains from fraudulent trading table and historical trading table, do not need to calculate, as dealing money, exchange hour etc.Context variable needs to obtain from transaction same card number, needs to carry out certain calculating or judgement, as transaction whether in areal, whether transaction in same trade company etc.Statistical variable is this card number or this card number transaction statistical information of trade company within a period of time, as in 30 days with the average dealing money of card number every, card to be concluded the business average every day such as stroke count etc.
3) data prediction
A. the characteristic variable of sampling sample is calculated;
B. variable discretize, carries out sliding-model control for the continuous variable in sample data, the variablees such as such as dealing money;
C. carry out randomly ordered to sample data.
4) generate sample data and and then generate Random Forest model
As shown in Figure 2, the training step of model is as follows for the product process figure of Random Forest model:
A. sample random sampling
The pretreated sample of tentation data is S, samples k time with putting back at random, and each sample size is 2/3rds of sample S, and sampling sample set is { s
1..., s
k.
B. selected characteristic variable
Suppose that characteristic variable adds up to N, for each increment originally chooses n characteristic variable, wherein
the variable chosen each time is as far as possible not identical.
C. decision-tree model is trained.
Every portion sampling sample training generates a decision-tree model, and symbiosis becomes k decision-tree model.Suppose that T is for a sampling sample, T=s
i, i=1 ..., k.Sample T comprises arm's length transaction, fraudulent trading two classifications, be defined as classified variable Y, value normal, fraud, wherein the quantity of training sample is Num (T), the quantity of arm's length transaction is Num (normal), and the quantity of fraudulent trading is Num (fraud).
● the Gini coefficient of calculation training sample.The Gini coefficient Gini (T) of training sample calculates by following formula.
Gini(T)＝1-p
normal(T)
2-p
fraud(T)
2
Wherein
represent the probability of arm's length transaction in training sample T,
represent that fraudulent trading is at the probability in training sample T.
● determine split vertexes.With set X={X
1..., X
n, represent original variable and the context variable of training sample, the value of each variable is Xi={c
i1..., c
im, suppose variable X i=c, c ∈ { c
i1..., c
imsample T is divided into two subset T (Xi≤c), T (Xi > c), calculate the Gini coefficient Gini (T this time divided
xi=c).
Wherein Num (T (Xi≤c)), Num (T (Xi > c)) are respectively the sample size of subset T (Xi≤c), T (Xi > c).Calculate the Gini coefficient that all variablees divide in all values, get Gini coefficient minimum be divided into best split vertexes.
● repeat previous step.Suppose Gini coefficient Gini (T '
xi=c) for minimum, then X
i=c is the best split vertexes of previous step, subset T (Xi≤c), T (Xi > c) repeats previous step respectively and determines next best split vertexes.Division is stopped for the moment: the classified variable Y of subset T (Xi≤c) or T (Xi > c) subset is same type when meeting the following conditions; Set X is without value in any case, and the Gini coefficient of division no longer reduces; Sample size on subset T (Xi≤c) or T (Xi > c) is less than threshold alpha.
D. decision-tree model assessment.
Supposing that T ' is for test sample book, is remaining 1/3rd samples of S, i.e. T '=S-T.Test sample book corresponding to each decision-tree model is respectively tested, and adds up the accuracy rate r of each decision-tree model classification
i, i=1 ..., k, if r
ibe less than threshold value beta, then this decision-tree model given up.Suppose that qualified decision-tree model quantity is l, corresponding accuracy rate is respectively r
i, i=1 ..., l.
E. the ballot weight of each decision tree is determined.
The classification results of random forest is the common decision of each decision tree ballot.Suppose that the output of each decision tree is y
i, i=1 ..., l, the weight of each decision tree is q
i, i=1 ..., l, q
iby following formulae discovery:
Each sample data is by the classification results of random forest, and namely the voting results RF of multiple decision tree is:
5) fraud detection is carried out to production transaction
The Random Forest model good by Training valuation carries out fraud detection to production transaction, generates fraud detection result set. and carry out manual confirmation to the transaction in fraud detection result set, the transaction confirming as swindle adds fraudulent trading table.Re-training Renewal model at set intervals, ensures the fraudulent trading rule that model identifiable design is up-to-date.
Above, composition graphs 1 and Fig. 2 particularly illustrate the financial transaction detection method based on random forest of the present invention.Those skilled in the art can understand, when without departing from the spirit and scope of the present invention, the financial transaction detection method based on random forest of the present invention can also be realized with corresponding hardware device, computer program or alternate manner.These change and replacement is interpreted as falling in claims of the present invention limited range.
Claims (18)
1., based on a financial transaction detection method for Random Forest model, comprising:
A () obtains historical trading table and fraudulent trading table;
B () utilizes described historical trading table and described fraudulent trading table to construct sample data sets, described sample data sets comprises sample characteristics variable;
C () extracts many increments originally with putting back at random from described sample data sets;
D () is the characteristic variable of each this Stochastic choice of the increment equal number in described many increment bases, to generate and this corresponding decision-tree model of this increment, and then generate Random Forest model;
E () trains and assesses each of the multiple decision-tree models in described Random Forest model, to obtain the accuracy rate of each decision-tree model;
F () determines the ballot weight q of each decision-tree model based on described accuracy rate
i; And
G () utilizes the multiple decision-tree models in described Random Forest model to export y to the response of inputted data of financial transaction
iand described ballot weight q
i, obtain voting results RF according to following formula and judge whether described financial transaction exists swindle:
Wherein, l is the quantity of decision-tree model.
2. the method for claim 1, also comprises:
H described transaction, when judging that described financial transaction exists swindle, adds in fraud detection result set by ();
I () confirms the transaction in described fraud detection result set, and the transaction confirming as swindle added in described fraudulent trading table; And
J () re-executes step (a) and (b).
3. the method for claim 1, wherein step (e) comprises further:
(e1) Gini coefficient that in calculation training sample, all variablees divide in all values;
(e2) get Gini coefficient minimum be divided into the first best splitting point; And
(e3) based on described first best splitting point, described training sample is divided, and (e1) and (e2) is repeated respectively to determine the second best splitting point to the training sample after dividing.
4. the method for claim 1, wherein described sample characteristics variable comprises original variable, context variable and statistical variable.
5. method as claimed in claim 4, wherein, described original variable is the dealing money and exchange hour that directly obtain from described fraudulent trading table and described historical trading table.
6. method as claimed in claim 4, wherein, described context variable be conclude the business whether in areal and transaction whether in same trade company.
7. method as claimed in claim 4, wherein, described statistical variable is certain card number or this card number transaction trade company statistical information within a period of time.
8. the quantity of the method for claim 1, wherein selected characteristic variable is n, and the relation of the total N of n and characteristic variable is as follows:
9., the method for claim 1, wherein when the accuracy rate of certain decision-tree model is lower than a threshold value, from described Random Forest model, give up this decision-tree model.
10., based on a financial transaction checkout equipment for Random Forest model, comprising:
For obtaining the device of historical trading table and fraudulent trading table;
For utilizing described historical trading table and described fraudulent trading table to construct the device of sample data sets, described sample data sets comprises sample characteristics variable;
Many increments device is originally extracted with putting back to for having at random from described sample data sets;
For the characteristic variable for each this Stochastic choice of the increment equal number in described many increment bases, to generate and this corresponding decision-tree model of this increment, and then generate the device of Random Forest model;
For training and assess each of the multiple decision-tree models in described Random Forest model, to obtain the device of the accuracy rate of each decision-tree model;
For determining the ballot weight q of each decision-tree model based on described accuracy rate
idevice; And
For utilizing the multiple decision-tree models in described Random Forest model, y is exported to the response of inputted data of financial transaction
iand described ballot weight q
i, obtain voting results RF according to following formula and judge whether described financial transaction exists the device of swindle:
Wherein, l is the quantity of decision-tree model.
11. equipment as claimed in claim 10, also comprise:
For when judging that described financial transaction exists swindle, described transaction is added the device in fraud detection result set;
For confirming the transaction in described fraud detection result set, and the transaction confirming as swindle is added the device in described fraudulent trading table.
12. equipment as claimed in claim 10, wherein, for training each of the multiple decision-tree models in described Random Forest model and assess, so that the device obtaining the accuracy rate of each decision-tree model is configured to perform following steps:
(e1) Gini coefficient that in calculation training sample, all variablees divide in all values;
(e2) get Gini coefficient minimum be divided into the first best splitting point; And
(e3) based on described first best splitting point, described training sample is divided, and (e1) and (e2) is repeated respectively to determine the second best splitting point to the training sample after dividing.
13. equipment as claimed in claim 10, wherein, described sample characteristics variable comprises original variable, context variable and statistical variable.
14. equipment as claimed in claim 10, wherein, described original variable is the dealing money and exchange hour that directly obtain from described fraudulent trading table and described historical trading table.
15. equipment as claimed in claim 10, wherein, described context variable be conclude the business whether in areal and transaction whether in same trade company.
16. equipment as claimed in claim 10, wherein, described statistical variable is certain card number or this card number transaction trade company statistical information within a period of time.
17. equipment as claimed in claim 10, wherein, the quantity of selected characteristic variable is n, and the relation of the total N of n and characteristic variable is as follows:
18. equipment as claimed in claim 10, also comprise: for when the accuracy rate of certain decision-tree model is lower than a threshold value, give up the device of this decision-tree model from described Random Forest model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410361193.3A CN105279691A (en) | 2014-07-25 | 2014-07-25 | Financial transaction detection method and equipment based on random forest model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410361193.3A CN105279691A (en) | 2014-07-25 | 2014-07-25 | Financial transaction detection method and equipment based on random forest model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105279691A true CN105279691A (en) | 2016-01-27 |
Family
ID=55148646
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410361193.3A Pending CN105279691A (en) | 2014-07-25 | 2014-07-25 | Financial transaction detection method and equipment based on random forest model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105279691A (en) |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529960A (en) * | 2016-11-07 | 2017-03-22 | ***股份有限公司 | Fraud transaction detection method for electronic transaction |
| CN106789912A (en) * | 2016-11-22 | 2017-05-31 | 清华大学 | Router data plane anomaly detection method based on classification regression tree |
| CN106897931A (en) * | 2016-06-12 | 2017-06-27 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of abnormal transaction data and device |
| CN107330785A (en) * | 2017-07-10 | 2017-11-07 | 广州市触通软件科技股份有限公司 | A kind of petty load system and method based on the intelligent air control of big data |
| CN107423871A (en) * | 2017-04-24 | 2017-12-01 | 成都知数科技有限公司 | Financial air control field multiple features fusion extracting method |
| CN107481019A (en) * | 2017-07-28 | 2017-12-15 | 上海携程商务有限公司 | Order fraud recognition methods, system, storage medium and electronic equipment |
| CN107563645A (en) * | 2017-09-04 | 2018-01-09 | 杭州云算信达数据技术有限公司 | A kind of Financial Risk Analysis method based on big data |
| CN108074179A (en) * | 2017-12-07 | 2018-05-25 | 深圳乐信软件技术有限公司 | Financial air control tactics configuring method, system, server and storage medium |
| CN108305166A (en) * | 2018-04-04 | 2018-07-20 | 淮阴师范学院 | A kind of negative dealing fraud method towards financial field |
| CN108510096A (en) * | 2017-02-24 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | Trade company's attrition prediction method, apparatus, equipment and storage medium |
| WO2018166457A1 (en) * | 2017-03-15 | 2018-09-20 | 阿里巴巴集团控股有限公司 | Neural network model training method and device, transaction behavior risk identification method and device |
| CN108665159A (en) * | 2018-05-09 | 2018-10-16 | 深圳壹账通智能科技有限公司 | A kind of methods of risk assessment, device, terminal device and storage medium |
| CN108964951A (en) * | 2017-05-19 | 2018-12-07 | 腾讯科技（深圳）有限公司 | A kind of method and server of warning information acquisition |
| CN109472610A (en) * | 2018-11-09 | 2019-03-15 | 福建省农村信用社联合社 | A kind of bank transaction is counter to cheat method and system, equipment and storage medium |
| CN109767314A (en) * | 2018-12-14 | 2019-05-17 | 深圳壹账通智能科技有限公司 | Trade company's risk management and control method, device, computer equipment and storage medium |
| CN109791679A (en) * | 2016-09-26 | 2019-05-21 | 哈曼国际工业有限公司 | The system and method for prediction for automobile guarantee fraud |
| CN110264342A (en) * | 2019-06-19 | 2019-09-20 | 深圳前海微众银行股份有限公司 | A kind of business audit method and device based on machine learning |
| CN110390526A (en) * | 2018-04-18 | 2019-10-29 | 苏宁易购集团股份有限公司 | A kind of network trading analysis method and system |
| CN110414845A (en) * | 2019-07-31 | 2019-11-05 | 阿里巴巴集团控股有限公司 | For the methods of risk assessment and device of target transaction |
| CN110634067A (en) * | 2019-09-25 | 2019-12-31 | 上海应用技术大学 | Bank account abnormal transaction identification method |
| TWI684151B (en) * | 2016-10-21 | 2020-02-01 | 大陸商中國銀聯股份有限公司 | Method and device for detecting illegal transaction |
| CN111401906A (en) * | 2020-03-05 | 2020-07-10 | 中国工商银行股份有限公司 | Transfer risk detection method and system |
| CN111553685A (en) * | 2020-04-28 | 2020-08-18 | 中国工商银行股份有限公司 | Method, device, electronic equipment and storage medium for determining transaction routing channel |
| CN112308466A (en) * | 2020-11-26 | 2021-02-02 | 东莞市盟大塑化科技有限公司 | Enterprise qualification auditing method and device, computer equipment and storage medium |
| CN112949954A (en) * | 2019-11-22 | 2021-06-11 | 张捷 | Method for establishing financial fraud recognition model based on recognition learning |
| CN113056753A (en) * | 2018-11-21 | 2021-06-29 | 贝宝公司 | Machine learning based on post-transaction data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102890803A (en) * | 2011-07-21 | 2013-01-23 | 阿里巴巴集团控股有限公司 | Method and device for determining abnormal transaction process of electronic commodity |
| CN103473231A (en) * | 2012-06-06 | 2013-12-25 | 深圳先进技术研究院 | Classifier building method and system |
| CN103530540A (en) * | 2013-09-27 | 2014-01-22 | 西安交通大学 | User identity attribute detection method based on man-machine interaction behavior characteristics |
| CN103678659A (en) * | 2013-12-24 | 2014-03-26 | 焦点科技股份有限公司 | E-commerce website cheat user identification method and system based on random forest algorithm |
-
2014
- 2014-07-25 CN CN201410361193.3A patent/CN105279691A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102890803A (en) * | 2011-07-21 | 2013-01-23 | 阿里巴巴集团控股有限公司 | Method and device for determining abnormal transaction process of electronic commodity |
| CN103473231A (en) * | 2012-06-06 | 2013-12-25 | 深圳先进技术研究院 | Classifier building method and system |
| CN103530540A (en) * | 2013-09-27 | 2014-01-22 | 西安交通大学 | User identity attribute detection method based on man-machine interaction behavior characteristics |
| CN103678659A (en) * | 2013-12-24 | 2014-03-26 | 焦点科技股份有限公司 | E-commerce website cheat user identification method and system based on random forest algorithm |
Non-Patent Citations (6)
| Title |
|---|
| J.BUREZ ET AL: "Handling class imbalance in customer churn prediction", 《EXPERT SYSTEM WITH APPLICATIONS》 * |
| 刘建丽等: "基于决策树的税务数据分析", 《现代计算机(专业版)》 * |
| 周丽峰: "基于非平衡数据分类的贷款违约预测研究", 《中国优秀硕士学位论文全文数据库 经济与管理科学辑》 * |
| 方匡南: "《随机森林组合预测理论及其在金融中的应用》", 31 May 2012, 厦门：厦门大学出版社 * |
| 方匡南等: "随机森林方法研究综述", 《统计与信息论坛》 * |
| 解明恩等: "《云南短期气候预测方法与模型》", 31 December 2000 * |
Cited By (34)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106897931A (en) * | 2016-06-12 | 2017-06-27 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of abnormal transaction data and device |
| CN109791679A (en) * | 2016-09-26 | 2019-05-21 | 哈曼国际工业有限公司 | The system and method for prediction for automobile guarantee fraud |
| TWI684151B (en) * | 2016-10-21 | 2020-02-01 | 大陸商中國銀聯股份有限公司 | Method and device for detecting illegal transaction |
| CN106529960A (en) * | 2016-11-07 | 2017-03-22 | ***股份有限公司 | Fraud transaction detection method for electronic transaction |
| CN106789912A (en) * | 2016-11-22 | 2017-05-31 | 清华大学 | Router data plane anomaly detection method based on classification regression tree |
| CN106789912B (en) * | 2016-11-22 | 2020-02-21 | 清华大学 | Router data plane abnormal behavior detection method based on classification regression decision tree |
| CN108510096A (en) * | 2017-02-24 | 2018-09-07 | 百度在线网络技术（北京）有限公司 | Trade company's attrition prediction method, apparatus, equipment and storage medium |
| CN108629413B (en) * | 2017-03-15 | 2020-06-16 | 创新先进技术有限公司 | Neural network model training and transaction behavior risk identification method and device |
| TWI689874B (en) * | 2017-03-15 | 2020-04-01 | 香港商阿里巴巴集團服務有限公司 | Method and device for neural network model training and transaction behavior risk identification |
| WO2018166457A1 (en) * | 2017-03-15 | 2018-09-20 | 阿里巴巴集团控股有限公司 | Neural network model training method and device, transaction behavior risk identification method and device |
| CN108629413A (en) * | 2017-03-15 | 2018-10-09 | 阿里巴巴集团控股有限公司 | Neural network model training, trading activity Risk Identification Method and device |
| CN107423871A (en) * | 2017-04-24 | 2017-12-01 | 成都知数科技有限公司 | Financial air control field multiple features fusion extracting method |
| CN108964951A (en) * | 2017-05-19 | 2018-12-07 | 腾讯科技（深圳）有限公司 | A kind of method and server of warning information acquisition |
| CN108964951B (en) * | 2017-05-19 | 2020-12-29 | 腾讯科技（深圳）有限公司 | Method for acquiring alarm information and server |
| CN107330785A (en) * | 2017-07-10 | 2017-11-07 | 广州市触通软件科技股份有限公司 | A kind of petty load system and method based on the intelligent air control of big data |
| CN107481019A (en) * | 2017-07-28 | 2017-12-15 | 上海携程商务有限公司 | Order fraud recognition methods, system, storage medium and electronic equipment |
| CN107563645A (en) * | 2017-09-04 | 2018-01-09 | 杭州云算信达数据技术有限公司 | A kind of Financial Risk Analysis method based on big data |
| CN108074179A (en) * | 2017-12-07 | 2018-05-25 | 深圳乐信软件技术有限公司 | Financial air control tactics configuring method, system, server and storage medium |
| CN108305166A (en) * | 2018-04-04 | 2018-07-20 | 淮阴师范学院 | A kind of negative dealing fraud method towards financial field |
| CN110390526A (en) * | 2018-04-18 | 2019-10-29 | 苏宁易购集团股份有限公司 | A kind of network trading analysis method and system |
| CN108665159A (en) * | 2018-05-09 | 2018-10-16 | 深圳壹账通智能科技有限公司 | A kind of methods of risk assessment, device, terminal device and storage medium |
| CN109472610A (en) * | 2018-11-09 | 2019-03-15 | 福建省农村信用社联合社 | A kind of bank transaction is counter to cheat method and system, equipment and storage medium |
| CN113056753A (en) * | 2018-11-21 | 2021-06-29 | 贝宝公司 | Machine learning based on post-transaction data |
| CN109767314A (en) * | 2018-12-14 | 2019-05-17 | 深圳壹账通智能科技有限公司 | Trade company's risk management and control method, device, computer equipment and storage medium |
| CN110264342A (en) * | 2019-06-19 | 2019-09-20 | 深圳前海微众银行股份有限公司 | A kind of business audit method and device based on machine learning |
| CN110264342B (en) * | 2019-06-19 | 2024-06-28 | 深圳前海微众银行股份有限公司 | Business auditing method and device based on machine learning |
| CN110414845A (en) * | 2019-07-31 | 2019-11-05 | 阿里巴巴集团控股有限公司 | For the methods of risk assessment and device of target transaction |
| CN110414845B (en) * | 2019-07-31 | 2023-09-19 | 创新先进技术有限公司 | Risk assessment method and device for target transaction |
| CN110634067A (en) * | 2019-09-25 | 2019-12-31 | 上海应用技术大学 | Bank account abnormal transaction identification method |
| CN112949954A (en) * | 2019-11-22 | 2021-06-11 | 张捷 | Method for establishing financial fraud recognition model based on recognition learning |
| CN112949954B (en) * | 2019-11-22 | 2023-11-07 | 张捷 | Method for establishing financial fraud recognition model based on recognition learning |
| CN111401906A (en) * | 2020-03-05 | 2020-07-10 | 中国工商银行股份有限公司 | Transfer risk detection method and system |
| CN111553685A (en) * | 2020-04-28 | 2020-08-18 | 中国工商银行股份有限公司 | Method, device, electronic equipment and storage medium for determining transaction routing channel |
| CN112308466A (en) * | 2020-11-26 | 2021-02-02 | 东莞市盟大塑化科技有限公司 | Enterprise qualification auditing method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105279691A (en) | Financial transaction detection method and equipment based on random forest model | |
| Martin et al. | Estimating the gravity model when zero trade flows are frequent and economically determined | |
| Dumitrescu* et al. | Backtesting value-at-risk: from dynamic quantile to dynamic binary tests | |
| CN100595780C (en) | Handwriting digital automatic identification method based on module neural network SN9701 rectangular array | |
| CN106408423A (en) | Method and system for risk assessment and method for constructing system for risk assessment | |
| CN110599336B (en) | Financial product purchase prediction method and system | |
| CN104572449A (en) | Automatic test method based on case library | |
| CN109635010B (en) | User characteristic and characteristic factor extraction and query method and system | |
| Plakandaras et al. | Do leading indicators forecast US recessions? A nonlinear re‐evaluation using historical data | |
| CN105354198A (en) | Data processing method and apparatus | |
| CN105528465A (en) | Credit status assessment method and device | |
| Dbouk et al. | Towards a machine learning approach for earnings manipulation detection | |
| CN110659961A (en) | Method and device for identifying off-line commercial tenant | |
| CN109508807A (en) | Lottery user liveness prediction technique, system and terminal device, storage medium | |
| CN109376179A (en) | A kind of sample equilibrating method in data mining | |
| EP3223227A1 (en) | System and method for predicting financial and / or market data | |
| CN103106329B (en) | A kind of training sample constructed in groups method for SVR short-term load forecasting | |
| CN104537383A (en) | Massive organizational structure data classification method and system based on particle swarm | |
| CN112434862B (en) | Method and device for predicting financial dilemma of marketing enterprises | |
| Zhang et al. | Research on personal credit scoring model based on multi-source data | |
| CN111046947A (en) | Training system and method of classifier and identification method of abnormal sample | |
| CN109992592A (en) | Impoverished College Studentss recognition methods based on campus consumption card pipelined data | |
| CN111160929B (en) | Method and device for determining client type | |
| Akinci et al. | Comparison of iron and steel production defects using classification algorithms | |
| CN109146549A (en) | Lottery user product participation prediction technique, system and equipment, storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160127 |
|
| RJ01 | Rejection of invention patent application after publication |