CN105005575A - Quick developing interface method for enterprise intelligent prediction - Google Patents
Quick developing interface method for enterprise intelligent prediction Download PDFInfo
- Publication number
- CN105005575A CN105005575A CN201510097796.1A CN201510097796A CN105005575A CN 105005575 A CN105005575 A CN 105005575A CN 201510097796 A CN201510097796 A CN 201510097796A CN 105005575 A CN105005575 A CN 105005575A
- Authority
- CN
- China
- Prior art keywords
- model
- modeling
- enterprise
- intelligent prediction
- enterprise intelligent
- 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 quick developing interface method for enterprise intelligent prediction. Due to the influence of tight coupling of an enterprise intelligent prediction modeling process and an intelligent prediction application on wide use of the data mining technology, a system architecture is needed to realize loose coupling between a service provider and a service consumer. The scheme of the method is as follows: firstly, a prediction model needed by an enterprise is constructed by use of an intelligent prediction modeling tool, secondly, the prediction model is released in the form of an XML file interface, and finally, a third-party application can invoke the model result to realize enterprise-class intelligent prediction application. According to the method, data mining services are provided for the third-party application in the form of an XML model interface file, and therefore, the use difficulty of data mining is greatly reduced and effective integration of data mining and third-party software is realized in such a manner at an extremely low coupling degree. Furthermore, dynamic binding of the service consumer to different service providers is realized, and therefore, the integration of the data mining services in data mining application can be realized.
Description
Technical field
The present invention relates to Enterprise Data digging technology application, be specifically related to a kind of enterprise and build forecast model and the method for quick calling model interface.
Background technology
Universal along with ecommerce, each large business web site usage data digging technology on a large scale, and therefrom obtain commercial value rapidly.Such as, domestic a lot of online shopping mall has brought into use data mining technology to carry out Customer clustering or commodity association popularization.In addition, the demand of search engine enterprise usage data digging technology is also very urgent.Technically, they need usage data mining algorithm to find association between Web page and structural relation, better carry out Web page push; From commercial angle, each large search engine needs obtain more advertising income, need click traffic data analysis, to realize maximum commercial profit.
Generally speaking, the data scale accumulated in enterprise is more and more huger.How to effectively utilize historical data, excavate the analytical information of value, thus help enterprise can make decision-making correct in time to change in future, be finally active in the market competition of fierceness, become the problem that current enterprise is more and more urgently wanted to solve.With abroad comparing, China due to the level of informatization not high, enterprises information is imperfect, and retail trade, bank, insurance, security etc. are very not desirable to the application of data mining.But along with the aggravation of market competition, every profession and trade more and more seems strong to the wish of data mining technology, can estimate, the coming years, the data analysis application of every profession and trade was bound to develop into large-scale data excavation application from traditional statistical study.This trend not only requires that the technology such as data warehouse modeling and data mining should be used as substantive popularization, and requires that data mining technology merges mutually with basic technology, carries out proprietary data excavation in conjunction with the existing various infosystem of enterprise.And the Data Mining Tools that medium-sized and small enterprises shortcoming is own, in the face of ever-increasing mass data, medium-sized and small enterprises informatization equipment originally can not be satisfied the demand.How when not increasing too many construction cost, the data mining of self is gone out advantage, and whether enterprise, all in deep thinking, hesitates and hovers between increasing input.These data are multifarious, in disorder and magnanimity, and these all propose new requirement to the technological frame of data mining software and application model.
Along with application and the development of technology, mutual more and more frequent in enterprise between different information systems, the framework of system must can adapt to the following requirement carrying out data integration and business integration of enterprise.In addition, Modeling of Data Mining process with excavate combining closely of applying and have impact on widely using of data mining application and Data Mining Tools, this loose couplings also needing a kind of architectural framework can realize between ISP and service consumer.
Summary of the invention
The invention provides a kind of enterprise intelligent prediction fast Development interface method, is based on standard x ML interface, and third-party application, by this interface model file, realizes the fast Development that Enterprise Data excavates application.
For achieving the above object, technical scheme of the present invention is:
A kind of enterprise intelligent prediction fast Development interface method, is characterized in that comprising the steps:
(1) create enterprise intelligent prediction scheme: according to enterprise intelligent forecast demand, create enterprise intelligent prediction scheme at Modeling of Data Mining platform;
(2) load the prediction scheme that modeling sample data activation creates, load the modeling sample data that pre-service is good;
(3) model construction: select prediction algorithm to carry out model construction in Modeling Platform;
(4) Model publish: the model of structure is issued with XML file;
(5) model calls: to the XML forecast model file issued, and can directly call for third-party application;
(6) model optimization and reconstruct: when forecast result of model does not reach anticipate accuracy, reconstruction model Issuance model again.
Described Modeling Platform is set to a standalone module, comprises the function of project management, expert's sample management, model training, modelling verification and Model publish.
Described Modeling Platform is integrated with conventional modeling algorithm, comprises naive Bayesian network, bayesian belief network, decision table, CART decision tree, ID3 decision tree, C4.5 decision tree, BP neural network, LM neural network, RBF neural, FNN neural network, ANFIS neural network, WNN neural network, linear regression, successive Regression, logistic regression, isotonic regression, AdaBoostM1 algorithm, KStar algorithm, SVM support vector machine and the classification of K-arest neighbors.
The detailed process of described step (3) is: using modeling sample data random selecting 80% as training set, and from Modeling Platform, select one or more algorithm to carry out model training, after having trained, is used for model testing with another 20% sample.
The detailed process of described step (4) is: through the model be up to the standards, and generates XML model file by selecting " Issuance model ".
Model in described step (5) calls the development environment supporting to comprise VB, VC, PB, Dephi, C# .NET and JAVA.
The reconstruction model of described step (6) and again Issuance model, refer to when forecast result of model does not reach anticipate accuracy, again in Modeling Platform, new modeling sample data are imported and reconstruction model, again Issuance model again after empirical tests is qualified, new model file only need be replaced original model file by Issuance model again.
First said method builds forecast model needed for enterprise by intelligent predicting modeling tool, is issued by forecast model afterwards with XML file interface, finally by third-party application calling model result, realizes the application of enterprise-level intelligent predicting.Utilize the method to provide second development interface, user by simple configuration, and supplements necessary business processing and just can go out large enterprise's level intelligent predicting analytical applications by fast custom.The present invention has the following advantages:
(1) based on standard x ML interface, demand adjustment and application lifting that business data excavates application system can be adapted on the one hand, the trend of future technical advances can be met on the other hand, for third party's data mining application integration provides standards service;
(2) by model interface mode for third-party application provides data mining service, thus greatly reduce the use difficulty of data mining, realize data mining and third party software effective integration in the mode that a kind of degree of coupling is extremely low; Achieve the ISP that service consumer dynamic binding is different, thus achieve the integration of data mining service in data mining application;
(3) wide accommodation of the present invention, can be widely used in financial circles, insurance, telecommunications industry, securities business, manufacturing industry, retail trade, the every profession and trades such as bio-pharmaceuticals.
Accompanying drawing explanation
Fig. 1 is the overall procedure of a kind of enterprise intelligent prediction fast Development interface method of the present invention.
Fig. 2 is modeling and the interface interchange relation of a kind of enterprise intelligent prediction fast Development interface method of the present invention.
Fig. 3 is the model construction flow process of a kind of enterprise intelligent prediction fast Development interface method of the present invention.
Fig. 4 is thief-proof electric leakage assessment indicator system.
Embodiment
Below in conjunction with accompanying drawing and BP neural network algorithm, technical scheme of the present invention is described in detail.
As depicted in figs. 1 and 2, a kind of enterprise intelligent prediction fast Development interface method of the present invention, comprise the steps: first to create enterprise intelligent prediction scheme in Modeling Platform, then by the ready modeling sample data importing program, prediction algorithm is selected to carry out model construction in Modeling Platform again, the model built is issued with XML file, and third-party application, by this interface model file, realizes the fast Development that Enterprise Data excavates application.
Modeling Platform is set to a standalone module, comprises the function of project management, expert's sample management, model training, modelling verification and Model publish.
Modeling Platform is integrated with conventional modeling algorithm, as naive Bayesian network, bayesian belief network, decision table, CART decision tree, ID3 decision tree, C4.5 decision tree, BP neural network, LM neural network, RBF neural, FNN neural network, ANFIS neural network, WNN neural network, linear regression, successive Regression, logistic regression, isotonic regression, AdaBoostM1 algorithm, KStar algorithm, SVM support vector machine and the classification of K-arest neighbors etc.
Concrete steps are as follows:
Step 1: create enterprise intelligent prediction scheme: according to enterprise intelligent forecast demand, creates prediction scheme in Modeling Platform;
Step 2: load modeling sample data: activate the prediction scheme created, loads the modeling sample data that pre-service is good;
Step 3: model construction: select BP neural network algorithm (or other algorithm) to carry out model training and evaluation model effect to selected sample data from Modeling Platform.In actual use, need to test the network trained, namely be input in the network trained by one group of test sample book data incomplete same with training sample, calculate its result obtained whether in the accuracy rating of regulation, modeling process as shown in Figure 3.
In order to the convenience introduced, first define vector sum variable below:
Input vector
;
Hidden layer input vector
;
Hidden layer output vector
;
Output layer input vector
;
Output layer output vector
;
Desired output vector
;
The connection weights in input layer and middle layer
;
The connection weights of hidden layer and output layer
;
The each neuronic threshold value of hidden layer
;
The each neuronic threshold value of output layer
;
Sample data number
Activation function
.
Described
step 3specific implementation step is as follows:
s3.1netinit is given
,
,
with
compose the random number in an interval (-1,1) respectively, specification error function
, given computational accuracy value
with maximum study number of times
;
s3.2random selecting
individual input amendment
and the desired output of correspondence
;
s3.3calculate each neuronic input of hidden layer
, then use
the each neuronic output of hidden layer is calculated with activation function
;
s3.4utilize network desired output vector
, the actual output of network
, error of calculation function is to each neuronic partial derivative of output layer
,
s3.5utilize hidden layer to the connection weights of output layer
, output layer
with the output of hidden layer
error of calculation function is to each neuronic partial derivative of hidden layer
;
s3.6utilize output layer each neuronic
neuronic output each with hidden layer
revise connection weights
and threshold value
,
In formula,
before representing adjustment,
after representing adjustment,
for learning rate, value between (0,1);
s3.7use hidden layer each neuronic
neuronic input each with input layer
revise connection weight and threshold value,
s3.8calculate global error
s3.9judge whether network error meets the demands to work as
or study number of times is greater than the maximum times of setting
, then algorithm is terminated.Otherwise the desired output of the next learning sample of random selecting and correspondence, turns back to the 3rd step, enters next round learning process.
Step 4: Model publish: for the model be up to the standards, generates XML model file by " Issuance model ".XML model file example is as follows:
<？ xml version="1.0" encoding="UTF-8"？>
<net>
<inNodePoint>5</inNodePoint>
<hiddenNodePoint>6</hiddenNodePoint>
<outNodePoint>1</outNodePoint>
<toHiddenFunc>purelin</toHiddenFunc>
<toOutFunc>tansig</toOutFunc>
<toHiddenWeight1> -0.1270 -4.6935 -6.3104 -12.4372 -6.4823</toHiddenWeight1>
<toHiddenWeight2> -1.5449 -3.8611 -5.6794 -9.4634 -4.8803</toHiddenWeight2>
<toHiddenWeight3> -1.4783 -6.0146 -8.8709 -15.1786 -7.3042</toHiddenWeight3>
<toHiddenWeight4> 0.8520 9.6811 13.8141 26.6334 13.7302</toHiddenWeight4>
<toHiddenWeight5> 1.4993 9.3171 12.5671 22.9414 11.7810</toHiddenWeight5>
<toHiddenWeight6> 0.1773 -4.4971 -6.3788 -9.7846 -5.2757</toHiddenWeight6>
<toHiddenThreshold> -0.5387
-0.8213
-2.5992
3.3246
3.1803
-1.5261</toHiddenThreshold>
<toOutWeight>25.3992 -1.7717 20.1824 18.4815 18.7763 -14.0123</toOutWeight>
<toOutThreshold>4.4975</toOutThreshold>
</net>
Step 5: model calls: after Model publish, according to the integration environment needs, calls by different development languages such as VB, VC, PB, Dephi, C# .NET, JAVA.Model calls very simple, below for model calls example:
y=myFunction（String p,String modelTdm）
In formula:
MyFunction-calling interface function name
P-mode input
ModelTdm-model file
Y-model prediction exports
Step 6: model optimization and reconstruct: after Model publish, when prediction effect does not reach anticipate accuracy, again new modeling sample data can be imported and reconstruction model in Modeling Platform, can Issuance model again after empirical tests is qualified, new model file only need be replaced original model file by Issuance model again.
Application experiment example
Mainly through regular visit, periodic check ammeter, user, traditional thief-proof electric leakage method reports that the means such as stealing are to find stealing or measuring apparatus fault.But this method is too strong to the dependence of people, and the target of grabbing surreptitiously leakage detection is indefinite.A lot of power supply administration utilizes metering abnormal alarm function and electric energy data query function to carry out the on-line monitoring work of user power utilization situation mainly through marketing inspection personnel, power utility check personnel and gage work personnel at present, by gathering the information such as electricity exception, the warning of load exception, terminal alarms, main website, line loss exception, set up Data Analysis Model, carry out the fault that Real-Time Monitoring is stolen drain conditions and found measuring apparatus.Electric current, voltage, load data situation etc. that before and after occurring according to alert event, client's stoichiometric point is relevant, build the multiplexing electric abnormality analytical model based on index weighting, realizes checking whether client exists stealing, transgression for using electricity and measuring apparatus fault etc.
The diagnostic method of thief-proof electric leakage above, although some information of multiplexing electric abnormality can be obtained, but due to terminal wrong report or garbage too many, cannot reach real fast accurately location steal the object of electric leakage suspicion user and metering fault, often make inspecting personnel at a loss as to what to do.And when adopting this Method Modeling, the determination of each input pointer weight of model needs the knowledge and experience with expert, has very large subjectivity, there is obvious defect, so implementation result is often not fully up to expectations.By the method that this patent provides, evaluation model can be gone out by rapid build, realize thief-proof electric leakage automatic diagnosis.
The method how applied patent of the present invention and provide is provided below and realizes thief-proof electric leakage diagnosis modeling and model calls.
model construction
Step1: build assessment indicator system, thief-proof electric leakage evaluation index mainly comprises electricity class index, load class index, line loss class index and warning class index etc., and assessment indicator system is specifically shown in Fig. 4.
1) electricity class index, for doing difference after daily power consumption moving average, statistics consecutive variations amount.
2) load class index, for Real-time Load moves the accumulative variable quantity of level values mean square deviation.
3) warning class index, mainly contains voltage phase shortage, electric sampling open-phase and electric current reversed polarity etc. to the relevant terminal alarms that surreptitiously leaks electricity, and mainly judges this few class reports to the police whether have generation here.
4) line loss class index, line loss is here day separated time line loss, and its computing method are with electricity class index.It is that circuit line loss belonging to it increases that user the most directly embodies after stealing electric leakage.
5) certain kinds index, steal electric leakage suspicion coefficient for what calculated upper one day, adding up by this index, can increase the suspicion coefficient stealing electric leakage user, because it is a continuous print process that user steals electric leakage.
Step2: build expert sample bank, carry out model construction, also need to prepare expert's sample data, expert's sample comprises input and output two parts, input data can from metering system extract obtain, export data then need with reference to history steal electric leakage customer case by professional analyze obtain.
Owing to embodying the aspects such as the index of stealing electric leakage behavior mainly contains electricity exception, load exception, terminal alarms, main website report to the police, line loss is abnormal, in order to cover various mode of surreptitiously leaking electricity as far as possible comprehensively, modeling sample comprises all large users over nearly 5 years and steals the normal users of electric leakage user and random selecting 8%.
Surreptitiously electric leakage start time and the end time of stealing electric leakage user characterize the material time node that it steals electric leakage, on these timing nodes, the index such as electricity, load, warning, line loss also has obvious changing features, therefore will comprise each 30 workaday data before and after material time node when sample data extracts.
Combining assessment index system, carries out pre-service to the sample data extracted, and builds expert sample bank.
Step3: model training and evaluation, after expert's sample data is ready to complete, model construction can be carried out, namely the accumulative historical data of surreptitiously leaking electricity in a large number of electric system is analyzed, the status classification of all kinds of multiplexing electric abnormality is arranged, and provide corresponding result of determination, finally utilize data mining technology by the many potential key factor of containing in these data, the fact and the valuable information abstraction such as to associate out, analyze the correlativity of stealing the same other factors of leakage events (as electricity, line loss etc.), and and then build intelligent diagnostics model.
Thief-proof electric leakage is evaluated and is usually realized by building classification forecast model, relatively more conventional classification forecast model has artificial neural network, decision tree, logistic regression etc., and the artificial neural network algorithm that this example adopts Modeling Platform to provide carries out structure and the evaluation of thief-proof electric leakage diagnostic model.
Step4: Issuance model, for the model be up to the standards, generates XML model file by " Issuance model " of Modeling Platform.
model calls
Thief-proof electric leakage diagnostic module can be resolved the XML model file issued, and realizes thief-proof electric leakage intelligent diagnostics.
Carry out automatic diagnosis to Guangdong Province's power supply branch office 2000 stoichiometric points historical data of 3 years from 2010-1-1 to 2012-12-31 by this model, matching rate reaches 99.325%, and rate of failing to report is no more than 5%.
Model after issuing, when prediction effect does not reach anticipate accuracy, can import new modeling sample data and reconstruction model and redistribute model again in Modeling Platform.
Claims (7)
1. an enterprise intelligent prediction fast Development interface method, is characterized in that comprising the steps:
(1) create enterprise intelligent prediction scheme: according to enterprise intelligent forecast demand, create enterprise intelligent prediction scheme at Modeling of Data Mining platform;
(2) load the prediction scheme that modeling sample data activation creates, load the modeling sample data that pre-service is good;
(3) model construction: select prediction algorithm to carry out model construction in Modeling Platform;
(4) Model publish: the model of structure is issued with XML file;
(5) model calls: to the XML forecast model file issued, and can directly call for third-party application;
(6) model optimization and reconstruct: when forecast result of model does not reach anticipate accuracy, reconstruction model Issuance model again.
2. a kind of enterprise intelligent prediction fast Development interface method according to claim 1, is characterized in that:
Described Modeling Platform is set to a standalone module, comprises the function of project management, expert's sample management, model training, modelling verification and Model publish.
3. a kind of enterprise intelligent prediction fast Development interface method according to claim 1, is characterized in that:
Described Modeling Platform is integrated with conventional modeling algorithm, comprises naive Bayesian network, bayesian belief network, decision table, CART decision tree, ID3 decision tree, C4.5 decision tree, BP neural network, LM neural network, RBF neural, FNN neural network, ANFIS neural network, WNN neural network, linear regression, successive Regression, logistic regression, isotonic regression, AdaBoostM1 algorithm, KStar algorithm, SVM support vector machine and the classification of K-arest neighbors.
4. a kind of enterprise intelligent prediction fast Development interface method according to claim 1, is characterized in that:
The detailed process of described step (3) is: using modeling sample data random selecting 80% as training set, and from Modeling Platform, select one or more algorithm to carry out model training, after having trained, is used for model testing with another 20% sample.
5. a kind of enterprise intelligent prediction fast Development interface method according to claim 1, is characterized in that:
The detailed process of described step (4) is: through the model be up to the standards, and generates XML model file by selecting " Issuance model ".
6. a kind of enterprise intelligent prediction fast Development interface method according to claim 1, is characterized in that:
Model in described step (5) calls the development environment supporting to comprise VB, VC, PB, Dephi, C# .NET and JAVA.
7. a kind of enterprise intelligent prediction fast Development interface method according to claim 1, is characterized in that:
The reconstruction model of described step (6) and again Issuance model, refer to when forecast result of model does not reach anticipate accuracy, again in Modeling Platform, new modeling sample data are imported and reconstruction model, again Issuance model again after empirical tests is qualified, new model file only need be replaced original model file by Issuance model again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510097796.1A CN105005575A (en) | 2015-03-05 | 2015-03-05 | Quick developing interface method for enterprise intelligent prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510097796.1A CN105005575A (en) | 2015-03-05 | 2015-03-05 | Quick developing interface method for enterprise intelligent prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105005575A true CN105005575A (en) | 2015-10-28 |
Family
ID=54378251
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510097796.1A Pending CN105005575A (en) | 2015-03-05 | 2015-03-05 | Quick developing interface method for enterprise intelligent prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105005575A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256691A (en) * | 2018-02-08 | 2018-07-06 | 成都智宝大数据科技有限公司 | Refund Probabilistic Prediction Model construction method and device |
| CN108510082A (en) * | 2018-03-27 | 2018-09-07 | 苏宁易购集团股份有限公司 | The method and device that machine learning model is handled |
| CN109615081A (en) * | 2018-09-26 | 2019-04-12 | 阿里巴巴集团控股有限公司 | A kind of Model forecast system and method |
| CN110163431A (en) * | 2019-05-13 | 2019-08-23 | 中船第九设计研究院工程有限公司 | A kind of cruise tourist demand prediction technique based on machine learning |
| CN111047190A (en) * | 2019-12-12 | 2020-04-21 | 广西电网有限责任公司 | Diversified business modeling framework system based on interactive learning technology |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050228767A1 (en) * | 2004-04-13 | 2005-10-13 | International Business Machines Corporation | Method, system and program product for developing a data model in a data mining system |
| CN101169798A (en) * | 2007-12-06 | 2008-04-30 | 中国电信股份有限公司 | Data excavation system and method |
| CN101620691A (en) * | 2008-06-30 | 2010-01-06 | 上海全成通信技术有限公司 | Automatic data mining platform in telecommunications industry |
| CN101667191A (en) * | 2009-09-04 | 2010-03-10 | 北京邮电大学 | Information sharing method based on P2P |
| CN101847100A (en) * | 2010-06-08 | 2010-09-29 | 用友软件股份有限公司 | Method for expanding software application and device |
| CN102360377A (en) * | 2011-10-12 | 2012-02-22 | 中国测绘科学研究院 | Spatial clustering mining PSE (Problem Solving Environments) system and construction method thereof |
| CN102929606A (en) * | 2012-10-09 | 2013-02-13 | 曙光信息产业(北京)有限公司 | Cloud-computing-based plug-in model of data mining system |
-
2015
- 2015-03-05 CN CN201510097796.1A patent/CN105005575A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050228767A1 (en) * | 2004-04-13 | 2005-10-13 | International Business Machines Corporation | Method, system and program product for developing a data model in a data mining system |
| CN101169798A (en) * | 2007-12-06 | 2008-04-30 | 中国电信股份有限公司 | Data excavation system and method |
| CN101620691A (en) * | 2008-06-30 | 2010-01-06 | 上海全成通信技术有限公司 | Automatic data mining platform in telecommunications industry |
| CN101667191A (en) * | 2009-09-04 | 2010-03-10 | 北京邮电大学 | Information sharing method based on P2P |
| CN101847100A (en) * | 2010-06-08 | 2010-09-29 | 用友软件股份有限公司 | Method for expanding software application and device |
| CN102360377A (en) * | 2011-10-12 | 2012-02-22 | 中国测绘科学研究院 | Spatial clustering mining PSE (Problem Solving Environments) system and construction method thereof |
| CN102929606A (en) * | 2012-10-09 | 2013-02-13 | 曙光信息产业(北京)有限公司 | Cloud-computing-based plug-in model of data mining system |
Non-Patent Citations (1)
| Title |
|---|
| 张良均等: "《数据挖掘：实用案例分析》", 30 June 2013, 机械工业出版社 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108256691A (en) * | 2018-02-08 | 2018-07-06 | 成都智宝大数据科技有限公司 | Refund Probabilistic Prediction Model construction method and device |
| CN108510082A (en) * | 2018-03-27 | 2018-09-07 | 苏宁易购集团股份有限公司 | The method and device that machine learning model is handled |
| CN108510082B (en) * | 2018-03-27 | 2022-11-11 | 苏宁易购集团股份有限公司 | Method and device for processing machine learning model |
| CN109615081A (en) * | 2018-09-26 | 2019-04-12 | 阿里巴巴集团控股有限公司 | A kind of Model forecast system and method |
| CN110163431A (en) * | 2019-05-13 | 2019-08-23 | 中船第九设计研究院工程有限公司 | A kind of cruise tourist demand prediction technique based on machine learning |
| CN110163431B (en) * | 2019-05-13 | 2023-04-18 | 中船第九设计研究院工程有限公司 | Cruise ship tourist demand prediction method based on machine learning |
| CN111047190A (en) * | 2019-12-12 | 2020-04-21 | 广西电网有限责任公司 | Diversified business modeling framework system based on interactive learning technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gu et al. | Performance of transportation network under perturbations: Reliability, vulnerability, and resilience | |
| CN110097297B (en) | Multi-dimensional electricity stealing situation intelligent sensing method, system, equipment and medium | |
| CN103257921B (en) | Improved random forest algorithm based system and method for software fault prediction | |
| CN102866321B (en) | Self-adaptive stealing-leakage prevention diagnosis method | |
| US8751436B2 (en) | Analyzing data quality | |
| CN106650797B (en) | Power distribution network electricity stealing suspicion user intelligent identification method based on integrated ELM | |
| AU2001255994B2 (en) | Method of business analysis | |
| CN108647891A (en) | Data exception classification, Reasons method and device | |
| Grbac et al. | A second replicated quantitative analysis of fault distributions in complex software systems | |
| CN106161138A (en) | A kind of intelligence automatic gauge method and device | |
| Kang et al. | Optimal meter placement for water distribution system state estimation | |
| KR101648272B1 (en) | Method for forecasting household water demand and detecting water leakage based on complex event processing and machine learning | |
| CN105005575A (en) | Quick developing interface method for enterprise intelligent prediction | |
| Kim et al. | Sensitivity of earned value schedule forecasting to S-curve patterns | |
| CN105573997A (en) | Method and device for determining electric larceny suspect user | |
| CN107633050A (en) | A kind of method that stealing probability is judged based on big data analysis electricity consumption behavior | |
| CN106716454A (en) | Utilizing machine learning to identify non-technical loss | |
| CN110334879A (en) | Power grid bus reactive load forecasting method and device | |
| CN107798615A (en) | Declaration form renews charge difficulty Forecasting Methodology and device | |
| CN105844351A (en) | Prosperity index prediction method for electric power consumption market | |
| Hampton et al. | A review of current analytical methods, modelling tools and development frameworks applicable for future retail electricity market design | |
| Elkharbotly et al. | Toward sustainable water: prediction of non-revenue water via artificial neural network and multiple linear regression modelling approach in Egypt | |
| Zhang et al. | Real-time burst detection based on multiple features of pressure data | |
| Shaghaghi et al. | Proposing a new optimized forecasting model for the failure rate of power distribution network thermal equipment for educational centers | |
| Masud Rana et al. | Contaminant spread forecasting and confirmatory sampling location identification in a water-distribution system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 510663 Guangdong city of Guangzhou Province Economic and Technological Development Zone No. 232 Ke Xue Cheng Guang Bao Lu 2 building 405B room Applicant after: Zhang Liangjun Address before: 510663 Guangdong city of Guangzhou Province Economic and Technological Development Zone Science 232 Xue Cheng Guang Bao Lu Building No. 3 Room 501 Applicant before: Zhang Liangjun |
|
| COR | Change of bibliographic data | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20151028 |
|
| WD01 | Invention patent application deemed withdrawn after publication |