CN110210973A - Insider trading recognition methods based on random forest and model-naive Bayesian - Google Patents
Insider trading recognition methods based on random forest and model-naive Bayesian Download PDFInfo
- Publication number
- CN110210973A CN110210973A CN201910472108.3A CN201910472108A CN110210973A CN 110210973 A CN110210973 A CN 110210973A CN 201910472108 A CN201910472108 A CN 201910472108A CN 110210973 A CN110210973 A CN 110210973A
- Authority
- CN
- China
- Prior art keywords
- insider trading
- model
- random forest
- insider
- bayesian
- 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
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000007637 random forest analysis Methods 0.000 title claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 42
- 230000002068 genetic effect Effects 0.000 claims abstract description 32
- 238000012360 testing method Methods 0.000 claims abstract description 22
- 238000010276 construction Methods 0.000 claims abstract description 6
- 238000005457 optimization Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 13
- 210000000349 chromosome Anatomy 0.000 claims description 3
- 230000035772 mutation Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000003066 decision tree Methods 0.000 description 8
- 230000008901 benefit Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000000977 initiatory effect Effects 0.000 description 2
- 235000015170 shellfish Nutrition 0.000 description 2
- 239000010749 BS 2869 Class C1 Substances 0.000 description 1
- 239000010750 BS 2869 Class C2 Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- JHIVVAPYMSGYDF-UHFFFAOYSA-N cyclohexanone Chemical compound O=C1CCCCC1 JHIVVAPYMSGYDF-UHFFFAOYSA-N 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The insider trading recognition methods based on random forest and model-naive Bayesian that the invention discloses a kind of, this method obtains the insider trading sample data set under the different time window phase, construction feature index set is screened using Random Forest model, according to the Bayesian recognition model of the characteristic index collection building insider trading filtered out, using Bayesian recognition model carry out insider trading identification, obtain whether the result of insider trading;Whether subsequent supervision verifying insider trading recognition result is correct, and is trained update to Bayesian recognition model according to recognition result.The present invention establishes stock insider trading identification model, realizes and accurately identifies to whether test target carries out insider trading;In conjunction with quasi-Newton method and genetic algorithm, make the parameter of Random Forest model quickly, be accurately optimized to optimal solution, the solution of optimal solution is small to the dependence of initial value;The present invention is easily achieved, and performance is stablized, and as sample data increases, robustness, accuracy can be further increased.
Description
Technical field
The invention belongs to Securities Market Regulation fields, and in particular to a kind of based on random forest and model-naive Bayesian
Insider trading recognition methods.
Background technique
The securities market insider trading behavior severe jamming normal operation of stock market especially compromises medium and small investors benefit
Benefit.In recent years, how to identify that insider trading has become the hot spot of academia's concern.
Naive Bayes Classifier is one of learning efficiency, the preferable classifier of classifying quality compared with other classifiers,
Its advantage is simple for algorithm logic, be easily achieved, algorithm performance is stablized etc..The basis of Naive Bayes Classification is probability inference,
The probability that each condition known occurs, completes reasoning and decision task.
Genetic algorithm is a kind of chess game optimization algorithm, its main analog mechanism of biogenetics and natural selection.It loses
Propagation algorithm has powerful ability of searching optimum, and initiating searches speed quickly, is solving the complicated and changeable optimization problem of state
Aspect is advantageous, but genetic algorithm later period speed of searching optimization is slower, and optimizing result is relatively inaccessible to ideal precision.
Random forest is the classifier comprising multiple decision trees, and the classification results of output are by setting output individually
As a result comprehensive descision not only overcomes the select permeability of quantity of parameters, possesses higher robustness, but also has and can assess
Feature importance, be less prone to over-fitting, with can quick and precisely handle mass data the advantages of.
Therefore, genetic algorithm is improved, study a kind of combination random forests algorithm, Naive Bayes Classifier it is interior
Curtain transaction identification method.
Summary of the invention
The technical problem of the invention normal operation of stock market that has been the insider trading behavior severe jamming of securities market, damage
Investors' interest, is not easy to be identified;The purpose of the present invention is being directed to this technical problem, provide a kind of based on random forest and Piao
The insider trading recognition methods of plain Bayesian model carries out the parameter of Random Forest model in conjunction with quasi-Newton method and genetic algorithm
After optimization, building optimal characteristics index set is screened using Random Forest model, then passes through building model-naive Bayesian verification
Whether certificate share of market, which occurs insider trading, quickly identifies.
The technical scheme is that the insider trading recognition methods based on random forest and model-naive Bayesian, specifically
Steps are as follows,
Step 1: obtaining the insider trading sample data set under the different event time window phase;
Step 2: construction feature index set is screened using Random Forest model;
Step 3: according to the Bayesian recognition model of the characteristic index collection building insider trading filtered out;
Step 4: obtaining test target and construct test target data set;
Step 5: using Bayesian recognition model carry out insider trading identification, obtain whether the result of insider trading;
Step 6: whether subsequent supervision verifying insider trading recognition result is correct;
Step 6.1: if subsequent supervision verifying insider trading recognition result is correct, thening follow the steps 8;
Step 6.2: if subsequent supervision verifying insider trading recognition result is incorrect, thening follow the steps 7;
Step 7: by test target data set and whether sample data set and training update shellfish is added in the result of insider trading
This identification model of leaf;
Step 8: judging whether there is next test target;
Step 8.1: if there is next test target, thening follow the steps 4;
Step 8.2: if terminating without next test target.
In the step 2, the parameter of the random forests algorithm uses the optimization method of quasi-Newton method combination genetic algorithm
It determining, the optimization method of quasi-Newton method combination genetic algorithm includes the following steps,
Step 1: determining population size N, crossover probability pc, mutation probability ps, genetic algorithm target fitness threshold value η;
Step 2: carrying out genetic algorithm iteration, obtain next-generation group;
Step 3: calculating current group average fitness η ';
Step 4: judging whether η ' < η;
Step 4.1: if η ' < η, thens follow the steps 5;
Step 4.2: if η ' < η is invalid, thening follow the steps 2;
Step 5: recording optimal chromosome values;
Step 6: the iteration result that genetic algorithm is obtained is as the initial value of quasi-Newton iteration method;
Step 7: carrying out quasi-Newton iteration method;
Step 8: judging whether to reach default precision;
Step 8.1: if not up to default precision, executes step 7;
Step 8.2: if reaching default precision, terminating.
In the step 2, the screening construction feature index set is carried out special using the Gini coefficient in Random Forest model
The different degree for levying index calculates, and chooses optimal characteristic index according to prominence score and combines.
The prominence score of variable indicates that Gini coefficient is indicated with GI with VIM, and the calculation formula of the GI of node m is as follows
K indicates K classification, p in formulamkIndicate ratio shared by classification k in node m.
Feature XjIn the importance of node mGI variable quantity i.e. before and after node m branch
GI in formulal、GIrRespectively indicate the GI of latter two new node of branch.
If feature XjThe node set occurred in decision tree is M, then feature XjIn i-th several importance
If the shared n tree of random forest, then
Prominence score is normalized
In formulaIt is characterized XjGini coefficient prominence score, c is characterized number.
Beneficial effects of the present invention:
1) present invention establishes stock insider trading identification model, realize to test target with the presence or absence of insider trading into
Row accurately identifies;
2) quasi-Newton method and genetic algorithm are combined, so that the parameter of Random Forest model is accurately optimized to optimal solution, most
The solution of excellent solution is small to the dependence of initial value;
3) optimal characteristic index combination is chosen using the Gini coefficient prominence score in Random Forest model, makes this hair
Bright insider trading recognition methods accuracy is good, high-efficient;
4) Naive Bayes Classifier is used, insider trading recognition methods of the invention is easily achieved, performance is stablized, and
With being continuously increased for sample data, robustness, the accuracy of insider trading recognition methods can be further increased.
Detailed description of the invention
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1 is the flow chart of the insider trading recognition methods based on random forest and model-naive Bayesian.
Fig. 2 is the flow chart of the optimization method of quasi-Newton method combination genetic algorithm.
Fig. 3 is Random Forest model schematic diagram.
Fig. 4 is the flow chart of model-naive Bayesian assorting process.
Fig. 5 is the flow chart of genetic algorithm.
Specific embodiment
As shown in Figure 1-3, the insider trading recognition methods based on random forest and model-naive Bayesian, specific steps are such as
Under,
Step 1: the corresponding difference of stock sample for the generation insider trading that stock supervisory committee announces is obtained using web crawlers
Characteristic index under event time window phase specifically includes securities market Microscopic Indexes, Corporate Finance index and company governance and refers to
Mark;Obtain the insider trading sample data set under the different event time window phase;
Step 2: the calculating of characteristic index different degree is carried out using the Gini coefficient of Random Forest model, according to prominence score
Choose optimal characteristic index combination;
Step 3: according to the Bayesian recognition model of the characteristic index collection building insider trading filtered out, by insider trading sample
Notebook data collection is as training dataset training Bayesian recognition model;
Step 4: obtaining test target and construct test target data set;
Step 5: using Bayesian recognition model carry out insider trading identification, obtain whether the result of insider trading;
Step 6: whether subsequent supervision verifying insider trading recognition result is correct;
Step 6.1: if subsequent supervision verifying insider trading recognition result is correct, thening follow the steps 8;
Step 6.2: if subsequent supervision verifying insider trading recognition result is incorrect, thening follow the steps 7;
Step 7: by test target data set and whether sample data set and training update shellfish is added in the result of insider trading
This identification model of leaf;
Step 8: judging whether there is next test target;
Step 8.1: if there is next test target, thening follow the steps 4;
Step 8.2: if terminating without next test target.
Characteristic index includes Company Financial index and company governance index disclosed in securities market personal share in step 1,
It further include by CAMP (Capital Asset Pricing Model), GARCH (Generalized AutoRegressive
Conditional Heteroskedasticity Model) model calculate personal share securities market Microscopic Indexes.
In step 2, the parameter of the random forests algorithm is determined using the optimization method of quasi-Newton method combination genetic algorithm,
The optimization method of quasi-Newton method combination genetic algorithm includes the following steps,
Step 1: determining population size N, crossover probability pc, mutation probability ps, genetic algorithm target fitness threshold value η;
Step 2: carrying out genetic algorithm iteration, obtain next-generation group;
Step 3: calculating current group average fitness η ';
Step 4: judging whether η ' < η;
Step 4.1: if η ' < η, thens follow the steps 5;
Step 4.2: if η ' < η is invalid, thening follow the steps 2;
Step 5: recording optimal chromosome values;
Step 6: the iteration result that genetic algorithm is obtained is as the initial value of quasi-Newton iteration method;
Step 7: carrying out quasi-Newton iteration method;
Step 8: judging whether to reach default precision;
Step 8.1: if not up to default precision, executes step 7;
Step 8.2: if reaching default precision, terminating.
The iterative process of quasi-Newton method are as follows:
xk+1=xk+λkdk
WhereinFor Newton direction,For gradient, λkFor along Newton direction
The step-length of linear search.
In step 2, the screening construction feature index set is chosen optimal characteristic index according to prominence score and is combined,
The prominence score of variable indicates that Gini value is indicated with GI with VIM, and the calculation formula of the GI of node m is as follows
K indicates K classification, p in formulamkIndicate ratio shared by classification k in node m.
Feature XjIn the importance of node mGI variable quantity i.e. before and after node m branch:
GI in formulal、GIrRespectively indicate the GI of latter two new node of branch.
If feature XjThe node set occurred in decision tree is M, then feature XjIn i-th several importance:
If the shared n tree of random forest, then
Then prominence score is normalized
In formulaIt is characterized XjGini coefficient prominence score, c is characterized number.
As shown in figure 3, the target of Random Forest model is by combining to obtain multiple weak learning machines such as single decision tree
One strong learning machine.Assuming that it is currently owned by N number of sample, characteristic M.Random Forest model uses bootstrap to adopt again first
Sample samples data set, training dataset of each N number of sample of stochastical sampling as single decision tree.In each node,
Algorithm randomly selects certain amount variable first, from finding the feature for being capable of providing optimal segmentation effect among them;Then, often
Decision tree all obtains a classification or prediction result.For classification problem, then the maximum class of probability value in prediction classification is selected
As final prediction.In random forests algorithm model, the present invention carries out parameter optimization using quasi-Newton method combination genetic algorithm.
Genetic algorithm has powerful ability of searching optimum, and initiating searches speed is fast, but later period speed of searching optimization is slower, and
Optimizing result is relatively inaccessible to ideal precision.Therefore, the present invention improves conventional genetic using quasi-Newton method combination genetic algorithm
The deficiency of algorithm, the iteration result that genetic algorithm is obtained is as the initial value of quasi-Newton method iteration, the flow chart of genetic algorithm
As shown in Figure 5.
Genetic algorithm is sentenced in optimization process close to the degree of optimal solution, and accordingly using fitness evaluation individual in population
Whether disconnected individual enters follow-on evolution.Can the selection of fitness directly affect the convergence rate of genetic algorithm and obtain
Optimal solution.The inverse that feature combines corresponding Classification and Identification accuracy is set as fitness function by the present invention.
As shown in figure 4, naive Bayesian insider trading identification process is as follows:
The data sample X of label whether one is given without insider trading, with n dimensional feature vector X={ x1,x2,…,xn}
It indicates, describes sample X respectively in n feature { A1,A2,…,AnOn characteristic value.Class C1Indicate insider trading, class C2It indicates
Without insider trading, sample X is distributed into class CiCondition is
P(Ci/ X) > P (Cj/X)(1≤j≤m,j≠i)
I.e. sample is class CiProbability be greater than sample be other classes probability.
According to Bayes' theorem:
P (X) refers to that any one data object meets the probability of sample X in formula, and for all classes, it is constant.P
(Ci) be any one data object be class CiProbability, P (C can be usedi)=si/ s is calculated, siIt is class CiMiddle number of training, s are
Training sample sum.
When label whether given sample insider trading, it is assumed that each characteristic value condition of reciprocity is independent, such P (X/Ci)
It calculates and uses formula
Probability P (xk/Ci) estimated by training sample, in the present invention, feature AkIt is continuous, it is assumed that characteristic value is obeyed high
This distribution.Therefore have
In formulaIt is characterized AkGauss specification density function,Respectively class in training sample
It Wei not CiFeature AkMean value, variance.
For test target collection E, to each class CiCalculate P (X/Ci)P(Ci).Sample E is assigned to class Ci, and if only if P
(Ci/ X) > P (Cj/X),1≤j≤m,j≠i。
In embodiment, the stock of generation insider trading between 2001 to 2017 years of stock supervisory committee's announcement is obtained by web crawlers
Sample with corresponding of the same trade, scale quite and do not occurred similar case white sample it is 335 total, wherein black sample 171
It is a, 164, white sample.
18 are had chosen in terms of subsidiary company equity structure, financial data, improvement system and securities market Microscopic respectively
A characteristic index, as shown in Table 1.
One characteristic index table of table
Quasi-Newton method combination genetic algorithm optimizes the parameter of Random Forest model, parameter optimization result such as two institute of table
Show.
Two Random Forest model parameter list of table
Parameter | Optimal value |
The variable number mtry of binary tree is used in node | 4 |
The quantity ntree set in random forest | 341 |
The minimum number nodesize of decision tree nodes | 9 |
The maximum number maxnodes of decision tree nodes | 17 |
In a kind of embodiment, in order to detect the identification effect of insider trading recognition methods of the invention on unknown stock sample
Collected data set is divided into training set and forecast set in the ratio of 8:2 by fruit.Using training set as learning data, training is simple
Bayesian recognition model.Then it is inputted using forecast set as the test of Bayesian recognition model, carries out the identification inspection of insider trading
It surveys.
As shown in Table 3, the insider trading totality recognition correct rate that event time window phase is 30 is 76.19%.
The insider trading recognition result table that three event time window phase of table is 30
As shown in Table 4, the insider trading totality recognition correct rate that event time window phase is 60 is 74.6%.
The insider trading recognition result table that four event time window phase of table is 60
As shown in Table 5, the insider trading totality recognition correct rate that event time window phase is 90 is 79.37%.
The insider trading recognition result table that five event time window phase of table is 90
Claims (3)
1. the insider trading recognition methods based on random forest and model-naive Bayesian, which is characterized in that specific step is as follows,
Step 1: obtaining the insider trading sample data set under the different event time window phase;
Step 2: construction feature index set is screened using Random Forest model;
Step 3: according to the Bayesian recognition model of the characteristic index collection building insider trading filtered out;
Step 4: obtaining test target and construct test target data set;
Step 5: using Bayesian recognition model carry out insider trading identification, obtain whether the result of insider trading;
Step 6: whether subsequent supervision verifying insider trading recognition result is correct;
Step 6.1: if subsequent supervision verifying insider trading recognition result is correct, thening follow the steps 8;
Step 6.2: if subsequent supervision verifying insider trading recognition result is incorrect, thening follow the steps 7;
Step 7: by test target data set and whether sample data set and training update Bayes is added in the result of insider trading
Identification model;
Step 8: judging whether there is next test target;
Step 8.1: if there is next test target, thening follow the steps 4;
Step 8.2: if terminating without next test target.
2. the insider trading recognition methods according to claim 1 based on random forest and model-naive Bayesian, special
Sign is, in step 2, the parameter of the random forests algorithm is determined using the optimization method of quasi-Newton method combination genetic algorithm,
The optimization method of quasi-Newton method combination genetic algorithm includes the following steps,
Step 1: determining population size N, crossover probability pc, mutation probability ps, genetic algorithm target fitness threshold value η;
Step 2: carrying out genetic algorithm iteration, obtain next-generation group;
Step 3: calculating current group average fitness η ';
Step 4: judging whether η ' < η;
Step 4.1: if η ' < η, thens follow the steps 5;
Step 4.2: if η ' < η is invalid, thening follow the steps 2;
Step 5: recording optimal chromosome values;
Step 6: the iteration result that genetic algorithm is obtained is as the initial value of quasi-Newton iteration method;
Step 7: carrying out quasi-Newton iteration method;
Step 8: judging whether to reach default precision;
Step 8.1: if not up to default precision, executes step 7;
Step 8.2: if reaching default precision, terminating.
3. the insider trading recognition methods according to claim 1 or 2 based on random forest and model-naive Bayesian,
It is characterized in that, in step 2, the screening construction feature index set carries out feature using the Gini coefficient in Random Forest model
The different degree of index calculates, and chooses optimal characteristic index according to prominence score and combines.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910472108.3A CN110210973A (en) | 2019-05-31 | 2019-05-31 | Insider trading recognition methods based on random forest and model-naive Bayesian |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910472108.3A CN110210973A (en) | 2019-05-31 | 2019-05-31 | Insider trading recognition methods based on random forest and model-naive Bayesian |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110210973A true CN110210973A (en) | 2019-09-06 |
Family
ID=67790087
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910472108.3A Pending CN110210973A (en) | 2019-05-31 | 2019-05-31 | Insider trading recognition methods based on random forest and model-naive Bayesian |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110210973A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111179077A (en) * | 2019-12-19 | 2020-05-19 | 成都数联铭品科技有限公司 | Method and system for identifying abnormal stock transaction |
CN111199419A (en) * | 2019-12-19 | 2020-05-26 | 成都数联铭品科技有限公司 | Method and system for identifying abnormal stock transaction |
CN111210269A (en) * | 2020-01-02 | 2020-05-29 | 平安科技(深圳)有限公司 | Object identification method based on big data, electronic device and storage medium |
CN111368076A (en) * | 2020-02-27 | 2020-07-03 | 中国地质大学(武汉) | Bernoulli naive Bayesian text classification method based on random forest |
CN112445844A (en) * | 2020-11-27 | 2021-03-05 | 重庆医药高等专科学校 | Financial data management control system of big data platform |
CN112990592A (en) * | 2021-03-26 | 2021-06-18 | 广东工业大学 | Shared vehicle fault prediction method and system |
CN113033674A (en) * | 2021-03-25 | 2021-06-25 | 安徽理工大学 | Apple multispectral image nondestructive testing method based on Bayesian optimization random forest algorithm |
-
2019
- 2019-05-31 CN CN201910472108.3A patent/CN110210973A/en active Pending
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111179077A (en) * | 2019-12-19 | 2020-05-19 | 成都数联铭品科技有限公司 | Method and system for identifying abnormal stock transaction |
CN111199419A (en) * | 2019-12-19 | 2020-05-26 | 成都数联铭品科技有限公司 | Method and system for identifying abnormal stock transaction |
CN111179077B (en) * | 2019-12-19 | 2023-09-12 | 成都数联铭品科技有限公司 | Stock abnormal transaction identification method and system |
CN111199419B (en) * | 2019-12-19 | 2023-09-15 | 成都数联铭品科技有限公司 | Stock abnormal transaction identification method and system |
CN111210269A (en) * | 2020-01-02 | 2020-05-29 | 平安科技(深圳)有限公司 | Object identification method based on big data, electronic device and storage medium |
CN111210269B (en) * | 2020-01-02 | 2020-09-18 | 平安科技(深圳)有限公司 | Object identification method based on big data, electronic device and storage medium |
CN111368076A (en) * | 2020-02-27 | 2020-07-03 | 中国地质大学(武汉) | Bernoulli naive Bayesian text classification method based on random forest |
CN111368076B (en) * | 2020-02-27 | 2023-04-07 | 中国地质大学(武汉) | Bernoulli naive Bayesian text classification method based on random forest |
CN112445844A (en) * | 2020-11-27 | 2021-03-05 | 重庆医药高等专科学校 | Financial data management control system of big data platform |
CN113033674A (en) * | 2021-03-25 | 2021-06-25 | 安徽理工大学 | Apple multispectral image nondestructive testing method based on Bayesian optimization random forest algorithm |
CN112990592A (en) * | 2021-03-26 | 2021-06-18 | 广东工业大学 | Shared vehicle fault prediction method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110210973A (en) | Insider trading recognition methods based on random forest and model-naive Bayesian | |
Xia et al. | A boosted decision tree approach using Bayesian hyper-parameter optimization for credit scoring | |
Liang et al. | The effect of feature selection on financial distress prediction | |
Song et al. | A hybrid evolutionary computation approach with its application for optimizing text document clustering | |
CN110059852A (en) | A kind of stock yield prediction technique based on improvement random forests algorithm | |
Yi et al. | An improved initialization center algorithm for K-means clustering | |
CN101923604A (en) | Classification method for weighted KNN oncogene expression profiles based on neighborhood rough set | |
CN110210974A (en) | A kind of insider trading discriminating conduct based on particle group optimizing Incremental support vector machine | |
CN112001788A (en) | Credit card default fraud identification method based on RF-DBSCAN algorithm | |
CN116821715A (en) | Artificial bee colony optimization clustering method based on semi-supervision constraint | |
Picek et al. | Plant recognition by AI: Deep neural nets, transformers, and kNN in deep embeddings | |
Yan et al. | A novel clustering algorithm based on fitness proportionate sharing | |
Zhang et al. | Research on borrower's credit classification of P2P network loan based on LightGBM algorithm | |
Fan et al. | An improved quantum clustering algorithm with weighted distance based on PSO and research on the prediction of electrical power demand | |
Yazdani et al. | Fuzzy classification method in credit risk | |
Xu et al. | Sample selection-based hierarchical extreme learning machine | |
Zhao et al. | Spectral–spatial classification of hyperspectral images using trilateral filter and stacked sparse autoencoder | |
Tian et al. | A new majority weighted minority oversampling technique for classification of imbalanced datasets | |
Maratkhan et al. | Financial forecasting using deep learning with an optimized trading strategy | |
Sun et al. | Forecasting day ahead spot electricity prices based on GASVM | |
CN108805162A (en) | A kind of saccharomycete multiple labeling feature selection approach and device based on particle group optimizing | |
Nourahmadi et al. | Portfolio Diversification Based on Clustering Analysis | |
M John et al. | Predicting House Prices using Machine Learning and LightGBM. | |
Mukkamala et al. | Model selection and feature ranking for financial distress classification | |
Zhang et al. | A Weighted KNN Algorithm Based on Entropy Method |
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: 20190906 |