WO2022227213A1 - 行业推荐方法、装置、计算机设备及存储介质 - Google Patents

行业推荐方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2022227213A1
WO2022227213A1 PCT/CN2021/097272 CN2021097272W WO2022227213A1 WO 2022227213 A1 WO2022227213 A1 WO 2022227213A1 CN 2021097272 W CN2021097272 W CN 2021097272W WO 2022227213 A1 WO2022227213 A1 WO 2022227213A1
Authority
WO
WIPO (PCT)
Prior art keywords
industry
industries
predicted
return
ranking
Prior art date
Application number
PCT/CN2021/097272
Other languages
English (en)
French (fr)
Inventor
段洪云
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2022227213A1 publication Critical patent/WO2022227213A1/zh

Links

Images

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
    • G06Q40/06—Asset management; Financial planning or analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Definitions

  • the present application relates to the field of information technology, and in particular, to an industry recommendation method, apparatus, computer equipment and storage medium.
  • a single prediction model is usually used to predict the development prospects of the industry, and the corresponding industries are recommended for users according to the prediction results.
  • a single prediction model can only mine the historical data information of the industry from a certain angle, and cannot fully mine more effective information in the historical data, and thus cannot judge and evaluate the development prospects of the industry from multiple angles, which leads to the industry
  • the recommendation accuracy is low, and it is difficult to provide investors with comprehensive, objective and effective advice.
  • the present application provides an industry recommendation method, device, computer equipment and storage medium, which mainly can predict the development prospect of the industry from multiple angles, mine more effective information in historical data, and improve the recommendation accuracy of the industry.
  • an industry recommendation method including:
  • the excess rate of return and the historical data are jointly input into a preset industry ranking prediction model for ranking prediction, and industry ranking results corresponding to a plurality of industries to be predicted are obtained;
  • a corresponding industry is recommended from a plurality of the industries to be predicted based on the industry ranking result.
  • an industry recommendation device comprising:
  • the acquisition unit is used to acquire historical data of multiple industries to be predicted
  • a first forecasting unit configured to input the historical data into a preset excess rate of return forecasting model for rate prediction, and obtain excess rates of return corresponding to a plurality of industries to be predicted;
  • a second prediction unit configured to jointly input the excess rate of return and the historical data into a preset industry ranking prediction model for ranking prediction, and obtain industry ranking results corresponding to a plurality of industries to be predicted;
  • a recommending unit configured to recommend a corresponding industry from a plurality of the industries to be predicted based on the industry ranking result.
  • a computer-readable storage medium on which computer-readable instructions are stored, and when the program is executed by a processor, the following steps are implemented:
  • the excess rate of return and the historical data are jointly input into a preset industry ranking prediction model for ranking prediction, and industry ranking results corresponding to a plurality of industries to be predicted are obtained;
  • a corresponding industry is recommended from a plurality of the industries to be predicted based on the industry ranking result.
  • a computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, the processor implementing the following steps when executing the program :
  • the excess rate of return and the historical data are jointly input into a preset industry ranking prediction model for ranking prediction, and industry ranking results corresponding to a plurality of industries to be predicted are obtained;
  • a corresponding industry is recommended from a plurality of the industries to be predicted based on the industry ranking result.
  • An industry recommendation method, device, computer equipment and storage medium provided by this application compared with the current method of using a single prediction model to predict the development prospects of the industry, and recommending the corresponding industry to users according to the prediction results, this application can Obtain historical data of a plurality of industries to be predicted, and input the historical data into a preset excess rate of return prediction model to predict the rate of return, and obtain the excess rate of return corresponding to a plurality of industries to be predicted. The excess rate of return and the historical data are jointly input into the preset industry ranking prediction model for ranking prediction, and a plurality of industry ranking results corresponding to the industries to be predicted are obtained.
  • the ranking of the industry to be predicted can be predicted, and the ranking of multiple industries to be predicted can be predicted from the perspective of excess rate of return, and then the development prospects of multiple industries to be predicted can be determined according to the predicted industry ranking results, and based on the industry
  • the ranking results recommend industries with better development prospects to users, thereby realizing the prediction of the development prospects of the industry from multiple perspectives, improving the recommendation accuracy of the industry, and providing investors with more comprehensive, objective and effective suggestions.
  • FIG. 1 shows a flowchart of an industry recommendation method provided by an embodiment of the present application
  • FIG. 2 shows a flowchart of another industry recommendation method provided by an embodiment of the present application
  • FIG. 3 shows a schematic structural diagram of an industry recommendation device provided by an embodiment of the present application
  • FIG. 4 shows a schematic structural diagram of another industry recommendation device provided by an embodiment of the present application.
  • FIG. 5 shows a schematic diagram of an entity structure of a computer device provided by an embodiment of the present application.
  • an embodiment of the present application provides an industry recommendation method, as shown in FIG. 1 , the method includes:
  • historical data can be single-dimensional historical data or multi-dimensional historical data.
  • Multi-dimensional historical data specifically includes business data, financial data, recruitment data, and welfare treatment data generated by the industry to be predicted in the past period of time. and public opinion data, etc.
  • the industry to be predicted is an industry that users consider investing in, and the number of industries to be predicted is at least two.
  • the ranking of multiple industries to be predicted the future development of the industry to be predicted can be determined, and then from Among the industries to be predicted, select industries with better development conditions and recommend them to users for investment. For example, users consider investing in education, real estate, and catering industries.
  • By predicting the ranking results of the above three industries determine the real estate industry for a period of time in the future.
  • the execution subject of the embodiment of the present application is an apparatus or device capable of performing industry recommendations, and may be specifically set on the client side or the server side.
  • the device side when a user needs to make an industry recommendation, he or she can select multiple industries to be predicted on the device side. Then you can select the industries to be predicted in the device as the real estate industry, education industry and catering industry, and then can generate an industry recommendation request. After receiving the industry recommendation request, the device side will collect the history of multiple industries to be predicted in the past period of time. data, and use the historical data to predict the rankings of multiple industries to be predicted in the future. Specifically, in the process of collecting historical data, in order to improve the prediction accuracy of industry rankings, you can collect the industry rankings to be predicted in different dimensions. Historical data, and according to the forecasted time period, statistics of historical data in different dimensions are performed, so that the ranking results of multiple industries to be predicted can be predicted according to the historical data in different dimensions after statistics.
  • the weekly order volume, total production, recruitment and employee salaries of the real estate industry, education industry and catering industry from 2001 to 2005 were collected respectively, and the ranking of the real estate industry, education industry and catering industry in the next week using the above data was collected.
  • Make predictions specifically, according to the weekly order volume, total production, recruitment and employee salaries in 2001-2005 and the number of weeks included in 2001-2005, calculate the average weekly order volume, total production, The number of recruits and employee salaries, and then the calculated weekly average order volume, the weekly average total generated total, the weekly average number of recruits, and the weekly average employee salary are used as input data to predict the real estate industry, the catering industry and the education industry in the next week.
  • Ranking results so as to determine industries with better development prospects and recommend them to users according to the ranking results.
  • the excess rate of return prediction model may specifically be an excess rate of return regression prediction model, or may be other types of prediction models, which are not specifically limited in the embodiment of the present application.
  • step 102 specifically includes: determining the weight parameters corresponding to the historical data in the different dimensions; The historical data under the dimension is added to obtain the excess rate of return corresponding to the multiple industries to be predicted.
  • the statistical historical data of the real estate industry is input into the preset excess return rate regression prediction model to predict the rate of return. Based on the weight values corresponding to the historical data in different dimensions in the prediction model, the The historical data is added to obtain the excess rate of return of the real estate industry in the next week. Similarly, the excess rate of return of the catering industry and the education industry in the next week can be obtained. Forecast the development prospects of the industry.
  • the preset industry ranking prediction model may specifically be a learning ranking model, or may be other types of prediction models, which are not specifically limited in the embodiments of the present application, and the specific process of ranking industries to be predicted is an optional implementation manner
  • Step 103 specifically includes: grouping a plurality of the industries to be predicted in pairs to obtain multiple groups of industries to be predicted; inputting the respective excess returns and historical data corresponding to the multiple groups of industries to be predicted into a preset industry ranking
  • the prediction model performs ranking prediction to obtain intra-group ranking results corresponding to the multiple groups of industries to be predicted; according to the intra-group ranking results, industry ranking results corresponding to a plurality of industries to be predicted are determined.
  • x 1 represents the weekly average order volume of the real estate industry
  • x 2 represents the real estate industry
  • x 3 represents the weekly average number of recruits in the real estate industry
  • x 4 represents the weekly average employee salary in the real estate industry
  • x 5 represents the excess rate of return of the real estate industry.
  • the input data corresponding to the catering industry can be obtained.
  • the specific prediction process is shown in step 204, so that not only can the historical The industry ranking can be predicted from the perspective of data, and the industry ranking can also be predicted from the perspective of excess rate of return, so that the industry ranking of the predicted industry can be predicted from multiple perspectives, so as to ensure the accuracy of the industry ranking results. Further, According to the sorting result, the industry development prospects of the industries to be predicted can be determined, so as to recommend industries with better development prospects to the user for investment.
  • the industries whose rankings meet the requirements of the preset conditions may be selected from a plurality of industries to be predicted according to the industry ranking results and recommended to the user.
  • the top industry is recommended to the user, or the industry that ranks within the preset ranking range among the industries to be predicted is recommended to the user, so that the user can invest according to the recommended industry.
  • the industry recommendation method provided by the embodiment of the present application compared with the current method of using a single prediction model to predict the development prospects of the industry, and recommending the corresponding industry to the user according to the prediction result, the present application can obtain multiple industries to be predicted. and input the historical data into the preset excess rate of return prediction model to predict the rate of return, and obtain a plurality of excess rates of return corresponding to the industries to be predicted. At the same time, the excess rate of return and all The above historical data are jointly input into the preset industry ranking prediction model for ranking prediction, and industry ranking results corresponding to multiple industries to be predicted are obtained, so that the ranking of multiple industries to be predicted can not only be predicted from the perspective of industry historical data.
  • the embodiment of the present application provides another industry recommendation method, as shown in FIG. 2 , the method includes:
  • the monthly average order quantity, the monthly maximum order quantity, the monthly minimum order quantity, the order quantity median, the monthly average production total, the monthly maximum production total, the monthly minimum production total, the monthly total production median, etc. can also be counted.
  • the data predicts the ranking of the industry to be predicted in the next month. It should be noted that because the shorter the forecast period, the higher the accuracy of the forecast will be, which is more conducive to providing effective advice to investors. Therefore, the to-be-predicted can be predicted on a weekly basis.
  • the ranking of the industry in the next consecutive weeks, and based on the ranking results, the development of the industry to be predicted in the next consecutive weeks can be determined, so as to give investors effective advice in the long run.
  • the preset excess rate of return prediction model may specifically be a multiple linear regression model, and the statistical historical data in different dimensions are input into the multiple linear regression model to predict the rate of return. Based on the weight parameters in the multiple linear regression model, the statistical The historical data in different dimensions are added to obtain the excess rate of return of the industry to be predicted in the future, so as to predict the development prospects of the industry to be predicted from the perspective of excess rate of return.
  • step 203 specifically includes: determining the industry rankings corresponding to the historical data in the different dimensions; inputting the industry rankings into a preset rate of return change trend prediction model for trend prediction, and obtaining Return rate change trends corresponding to a plurality of the industries to be predicted respectively.
  • the preset yield change trend prediction model may specifically be a logistic regression model, or a classification model such as a support vector machine, a decision tree, etc., and the yield change trend includes a relatively good yield trend, a general yield trend, and a poor yield trend.
  • the weekly average number of recruits and the average weekly employee salary in all industries and then input the industry rankings corresponding to the historical data in different dimensions of the real estate industry, education industry and catering industry into the preset logistic regression model.
  • Trend prediction obtain the probability value of the real estate industry, education industry and catering industry belonging to different yield trend, and then according to the probability value, determine the corresponding yield trend of the real estate industry, education industry and catering industry respectively.
  • the method further includes: performing anomaly detection on the historical data under the different dimensions, Determine the abnormal detection results corresponding to the historical data in the different dimensions, and at the same time, input the industry rankings into the preset rate of return change trend prediction model for trend prediction, and obtain a plurality of the industries to be predicted respectively.
  • the corresponding yield change trend includes: jointly inputting the abnormality detection result and the industry ranking into a preset yield change trend prediction model for trend prediction, and obtaining yield changes corresponding to a plurality of industries to be predicted respectively trend.
  • the input data corresponding to the real estate industry is (z 1 , z 2 , z 3 , z 4 , z 5 , z 6 , z 7 , z 8 ), where z 1 , z 2 , z 3 , z 4 , z 5 represent the weekly average orders of the real estate industry during 2001-2005, respectively z 5 , z 6 , z 7 , and z 8 represent outliers corresponding to order volume, total generation, number of recruits, and employee salaries, respectively, and take them as input
  • the data is input into the preset regression model for trend prediction, and the corresponding yield trend of the real estate industry is obtained.
  • the prediction process of the yield change trend can be realized by two logistic regression models, that is, the yield change trend prediction model includes the positive change trend of the yield rate.
  • the yield change trend prediction model includes the positive change trend of the yield rate.
  • a prediction model and a negative change trend prediction model of yield based on which, the industry ranking is input into the preset yield change trend prediction model for trend prediction, and the respective yields corresponding to a plurality of industries to be predicted are obtained.
  • Change trends including: inputting the industry rankings into a preset positive change trend prediction model of yields for trend prediction, and obtaining multiple prediction results of the forward trend of returns corresponding to the industries to be predicted;
  • the ranking is input into a preset negative change trend prediction model of yield for trend prediction, and a plurality of negative yield trend prediction results corresponding to the industries to be predicted are obtained.
  • the preset rate of return positive change trend prediction model can be a positive logistic regression model
  • the corresponding prediction results of the positive logistic regression model include a good rate of return trend and a moderate rate of return trend
  • the model can be a negative logistic regression model
  • the prediction results corresponding to the negative logistic regression model include a general return trend and a poor return trend.
  • the industry rankings and outliers corresponding to the historical data in different dimensions are input into the preset forward logistic regression model for trend prediction, and the positive trend prediction result of the rate of return of the industry to be predicted is obtained.
  • the industry rankings and outliers corresponding to the historical data under the dimension are input into the preset negative logistic regression model for trend prediction, and the negative trend prediction result of the yield of the industry to be predicted is obtained.
  • the forecast results of the rate of return, the positive trend of the rate of return and the forecast result of the negative trend of the rate of return are used to forecast the development prospects of the forecasting industry.
  • step 204 specifically includes: jointly inputting the prediction result of the positive trend of the yield rate, the prediction result of the negative trend of the yield rate, the excess rate of return and the historical data into the preset industry ranking prediction
  • the model performs ranking prediction, and obtains industry ranking results corresponding to a plurality of the industries to be predicted respectively.
  • the preset industry ranking prediction model may specifically be the RankNet ranking model.
  • the industry to be predicted is first divided into two groups to obtain a plurality of industry groups, and then each industry group is obtained separately.
  • the positive trend prediction results, negative yield trend prediction results, excess rate of return and historical data corresponding to the group of industries are input into the RankNet ranking model for prediction, and the ranking results between each group of industries are obtained, and then according to each group of industries The ranking results among all industries to be predicted are determined.
  • the industries to be predicted include industry A, industry B and industry C, and the industries to be predicted are combined in pairs to obtain industry A and industry B, industry B and industry C, industry A and industry C, and then the corresponding rate of return of each group of industries is calculated.
  • the positive trend prediction results, negative yield trend prediction results, excess returns and historical data are input into the RankNet ranking model for prediction.
  • the RankNet ranking model gives the classification label 1 Or 0, that is, the RankNet sorting model is actually a binary classifier at this time, which can output the probability value that the industry A to be predicted has a better development prospect than the industry B to be predicted, and the probability value that the industry B to be predicted has a better development prospect than the industry A to be predicted.
  • the classification label between industry A to be predicted and industry B to be predicted can be determined, and according to the classification label, the order between industry A to be predicted and industry B to be tested can be determined.
  • the output classification label is 1. It is determined that industry A to be predicted is arranged before industry B to be predicted.
  • the order of arrangement between industry B and industry C, as well as the order of industry A and industry C can be determined, and then according to the obtained industry corresponding to each group
  • the arrangement order of industry A, industry B and industry C can be determined, that is, the industry ranking corresponding to industry A, industry B and industry C can be determined.
  • the forecast results of the positive trend of the rate of return, the forecast result of the negative trend of the rate of return, the excess rate of return and the historical data corresponding to the sample industries can be collected in advance, and the sample industries can be divided into groups. , according to the development of the sample industry in the past period of time, mark each group of sample industries, use the labeled sample industries as the training set, and build a RankNet sorting model by training the training set. as follows:
  • the ith industry in the industry set S to be predicted is denoted as U i
  • its corresponding feature vector is denoted as xi
  • the RankNet ranking model uses a fully connected layer to classify the input industry
  • C i,j is the metric loss function, It represents the true probability value, which is written as:
  • the function optimizes the parameter ⁇ to build the RankNet ranking model.
  • the industries ranked first among multiple industries to be predicted may be recommended to users for investment, or industries ranked within a preset range may be recommended to users for reference. For example, there are 10 industries to be predicted in total, The top 3 industries are recommended to users for reference.
  • Another industry recommendation method provided by the embodiment of the present application compared with the current method of using a single prediction model to predict the development prospect of the industry, and recommending the corresponding industry to the user according to the prediction result, the present application can obtain multiple to-be-predicted
  • the historical data of the industry is input, and the historical data is input into the preset excess rate of return prediction model to predict the rate of return, and the excess rate of return corresponding to a plurality of industries to be predicted is obtained. At the same time, the excess rate of return is calculated.
  • Prediction can also predict the ranking of multiple industries to be predicted from the perspective of excess rate of return, and then can determine the development prospects of multiple industries to be predicted according to the predicted industry ranking results, and recommend to users based on the industry ranking results. For industries with better development prospects, it is possible to predict the development prospects of the industry from multiple perspectives, improve the recommendation accuracy of the industry, and provide investors with more comprehensive, objective and effective suggestions.
  • an embodiment of the present application provides an industry recommendation device.
  • the device includes: an acquisition unit 31 , a first prediction unit 32 , a second prediction unit 33 , and a recommendation unit 31 . unit 34.
  • the obtaining unit 31 is configured to obtain historical data of a plurality of industries to be predicted.
  • the obtaining unit 31 is a main functional module in the device for obtaining historical data of a plurality of industries to be predicted.
  • the first prediction unit 32 is configured to input the historical data into a preset excess rate of return prediction model to predict the rate of return, and obtain excess rate of return corresponding to a plurality of industries to be predicted.
  • the first predicting unit 32 is the main functional module in the device for inputting the historical data into a preset excess rate of return prediction model for rate prediction, and obtaining excess rate of return corresponding to a plurality of industries to be predicted, and is also the main function module. core module.
  • the second prediction unit 33 is configured to jointly input the excess rate of return and the historical data into a preset industry ranking prediction model for ranking prediction, and obtain industry ranking results corresponding to a plurality of industries to be predicted.
  • the second prediction unit 33 is a device that inputs the excess rate of return and the historical data together into a preset industry ranking prediction model for ranking prediction, and obtains industry ranking results corresponding to a plurality of industries to be predicted.
  • the main function module is also the core module.
  • the recommending unit 34 is configured to recommend a corresponding industry from a plurality of industries to be predicted based on the industry ranking result.
  • the recommending unit 34 is a main functional module in the device for recommending a corresponding industry from a plurality of the industries to be predicted based on the industry ranking result.
  • the first forecasting unit 32 is further configured to input the historical data into a preset yield change trend forecasting model for trend forecasting, and obtain a plurality of the The change trend of the yield corresponding to the industry to be forecasted.
  • the second prediction unit 33 is specifically set to input the change trend of the rate of return, the excess rate of return and the historical data into a preset industry ranking prediction model for ranking prediction, and obtain a plurality of the industries to be predicted. The corresponding industry ranking results.
  • the historical data is historical data in different dimensions.
  • the first prediction unit 32 includes: a determination module 321 and a prediction module 322 .
  • the determining module 321 is configured to determine the industry rankings corresponding to the historical data under the different dimensions.
  • the prediction module 322 is configured to input the industry rankings into a preset yield change trend prediction model for trend prediction, and obtain yield change trends corresponding to a plurality of industries to be predicted respectively.
  • the first prediction unit 32 further includes: a detection module 323 .
  • the detection module 323 is configured to perform abnormality detection on the historical data in the different dimensions, and determine abnormality detection results corresponding to the historical data in the different dimensions.
  • the prediction module 322 is specifically configured to input the abnormality detection result and the industry ranking together into a preset yield change trend prediction model for trend prediction, and obtain the yield changes corresponding to a plurality of industries to be predicted respectively. trend.
  • the yield change trend prediction model includes a positive yield change trend prediction model and a yield negative change trend prediction model
  • the prediction module 322 is further configured to input the industry ranking into a The predetermined trend prediction model of the positive rate of change performs trend forecasting, and obtains the forecast results of the positive rate of return corresponding to a plurality of the industries to be forecasted respectively.
  • the prediction module 322 is further configured to input the industry rankings into a preset negative change trend prediction model of yield for trend prediction, and obtain negative yield trend prediction results corresponding to a plurality of industries to be predicted respectively.
  • the second prediction unit 33 is specifically set to input the prediction result of the positive trend of the yield, the prediction result of the negative trend of the yield, the excess rate of return and the historical data into the preset industry ranking prediction.
  • the model performs ranking prediction, and obtains industry ranking results corresponding to a plurality of the industries to be predicted respectively.
  • the second prediction unit 33 in order to determine the industry ranking result corresponding to the industry to be predicted, includes: a grouping module 331 , a prediction module 332 and a determination module 333 .
  • the grouping module 331 is configured to group a plurality of the industries to be predicted in pairs to obtain multiple groups of industries to be predicted.
  • the prediction module 332 is configured to jointly input the excess rate of return and historical data corresponding to the multiple groups of industries to be predicted into the preset industry ranking prediction model for ranking prediction, and obtain the corresponding intra-group predictions of the multiple groups of industries to be predicted. Ranking results.
  • the determining module 333 is configured to determine industry ranking results corresponding to a plurality of industries to be predicted according to the ranking results within the group.
  • the first prediction unit 32 further includes an addition module 324 .
  • the determining module 321 is further configured to determine weight parameters corresponding to the historical data in the different dimensions.
  • the adding module 324 is configured to add historical data in different dimensions based on the weight parameter to obtain excess returns corresponding to a plurality of industries to be predicted.
  • an embodiment of the present application also provides a computer-readable storage medium on which computer-readable instructions are stored, and when the program is executed by a processor, the following steps are implemented: acquiring multiple The historical data of the industry to be predicted; input the historical data into the preset excess rate of return prediction model to predict the rate of return, and obtain a plurality of excess rates of return corresponding to the industries to be predicted; The historical data is jointly input into a preset industry ranking prediction model for ranking prediction, and industry ranking results corresponding to a plurality of industries to be predicted are obtained; and corresponding industries are recommended from a plurality of industries to be predicted based on the industry ranking results.
  • the computer-readable storage medium may be non-volatile or volatile.
  • an embodiment of the present application further provides a physical structure diagram of a computer device.
  • the computer device includes: a processor 41 , Memory 42, and computer-readable instructions stored on the memory 42 and executable on the processor, wherein both the memory 42 and the processor 41 are provided on the bus 43 when the processor 41 executes the program to achieve the following steps: obtaining The historical data of multiple industries to be predicted; input the historical data into a preset excess rate of return prediction model to predict the rate of return, and obtain the excess rate of return corresponding to a plurality of industries to be predicted; The historical data are jointly input into a preset industry ranking prediction model for ranking prediction, and industry ranking results corresponding to a plurality of industries to be predicted are obtained; and corresponding industry ranking results are recommended from a plurality of industries to be predicted based on the industry ranking results. industry.
  • the present application can obtain historical data of a plurality of industries to be predicted, input the historical data into a preset excess rate of return prediction model to predict the rate of return, and obtain a plurality of industries to be predicted corresponding to the respective industries.
  • the excess rate of return and the historical data are jointly input into the preset industry ranking prediction model for ranking prediction, and a plurality of industry ranking results corresponding to the industries to be predicted are obtained. It can predict the ranking of multiple industries to be predicted from the perspective of industry historical data, and can also predict the ranking of multiple industries to be predicted from the perspective of excess return, and then can determine the ranking of multiple industries to be predicted according to the predicted industry ranking results.
  • modules or steps of the present application can be implemented by a general-purpose computing device, and they can be centralized on a single computing device, or distributed in a network composed of multiple computing devices Alternatively, they may be implemented in program code executable by a computing device, such that they may be stored in a storage device and executed by the computing device, and in some cases, in a different order than here
  • the steps shown or described are performed either by fabricating them separately into individual integrated circuit modules, or by fabricating multiple modules or steps of them into a single integrated circuit module.
  • the present application is not limited to any particular combination of hardware and software.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Development Economics (AREA)
  • Accounting & Taxation (AREA)
  • Operations Research (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • General Engineering & Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Finance (AREA)
  • Game Theory and Decision Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Artificial Intelligence (AREA)
  • Technology Law (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请公开了一种行业推荐方法、装置、计算机设备及存储介质，涉及信息技术领域。其中方法包括：获取多个待预测行业的历史数据；将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；基于所述行业排名结果从多个所述待预测行业中推荐相应行业。本申请适用于行业推荐。

Description

行业推荐方法、装置、计算机设备及存储介质
本申请要求与2021年4月25日提交中国专利局、申请号为202110447074.X申请名称为“行业推荐方法、装置、计算机设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在申请中。
技术领域
本申请涉及信息技术领域，尤其是涉及一种行业推荐方法、装置、计算机设备及存储介质。
背景技术
随着经济的不断发展，涌现出了很多新行业，同时也吸引了大量投资者，投资者在进行投资之前会对即将投资行业的前景进行预测和评估，以便能够为自己带来可观的收益。
目前，在为投资者进行行业推荐时，通常采用单一的预测模型对行业的发展前景进行预测，并根据预测结果为用户推荐相应的行业。然而，单一的预测模型只能从某个角度挖掘行业的历史数据信息，无法全面地挖掘历史数据中更多的有效信息，进而无法从多个角度对行业的发展前景进行判断和评估，导致行业推荐的精度较低，难以为投资者提供全面、客观和有效的建议。
发明内容
本申请提供了一种行业推荐方法、装置、计算机设备及存储介质，主要在于能够从多个角度对行业的发展前景进行预测，挖掘历史数据中更多的有效信息，提高了行业的推荐精度。
根据本申请的第一个方面，提供一种行业推荐方法，包括：
获取多个待预测行业的历史数据；
将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
根据本申请的第二个方面，提供一种行业推荐装置，包括：
获取单元，用于获取多个待预测行业的历史数据；
第一预测单元，用于将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
第二预测单元，用于将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
推荐单元，用于基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
根据本申请的第三个方面，提供一种计算机可读存储介质，其上存储有计算机可读指令，该程序被处理器执行时实现以下步骤：
获取多个待预测行业的历史数据；
将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
根据本申请的第四个方面，提供一种计算机设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机可读指令，所述处理器执行所述程序时实现以下步骤：
获取多个待预测行业的历史数据；
将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
本申请提供的一种行业推荐方法、装置、计算机设备及存储介质，与目前采用单一的预测模型对行业的发展前景进行预测，并根据预测结果为用户推荐相应行业的方式相比，本申请能够获取多个待预测行业的历史数据，并将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业 分别对应的超额收益率，与此同时，将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，从而不仅能够从行业历史数据的角度对多个待预测行业的排名进行预测，还能够从超额收益率的角度对多个待预测行业的排名进行预测，进而能够根据预测的行业排名结果，确定多个待预测行业的发展前景，并基于该行业排名结果向用户推荐发展前景较好的行业，由此实现了从多个角度对行业的发展前景进行预测，提高了行业的推荐精度，能够为投资者提供更加全面、客观和有效的建议。
附图说明
此处所说明的附图用来提供对本申请的进一步理解，构成本申请的一部分，本申请的示意性实施例及其说明用于解释本申请，并不构成对本申请的不当限定。在附图中：
图1示出了本申请实施例提供的一种行业推荐方法流程图；
图2示出了本申请实施例提供的另一种行业推荐方法流程图；
图3示出了本申请实施例提供的一种行业推荐装置的结构示意图；
图4示出了本申请实施例提供的另一种行业推荐装置的结构示意图；
图5示出了本申请实施例提供的一种计算机设备的实体结构示意图。
具体实施方式
下文中将参考附图并结合实施例来详细说明本申请。需要说明的是，在不冲突的情况下，本申请中的实施例及实施例中的特征可以相互组合。
目前，单一的预测模型无法从多个角度对行业的发展前景进行判断和评估，导致行业推荐的精度较低，难以为投资者提供全面、客观和有效的建议。
为了解决上述问题，本申请实施例提供了一种行业推荐方法，如图1所示，所述方法包括：
101、获取多个待预测行业的历史数据。
其中，历史数据可以为单一维度的历史数据，也可以为多维度的历史数据，多维度的历史数据具体包括待预测行业在过去一段时间内生成的经营数据、财务 数据、招聘数据、福利待遇数据和舆情数据等，待预测行业为用户考虑进行投资的行业，待预测行业的数量至少为两个，通过对多个待预测行业的排名进行预测，能够确定待测行业的未来发展情况，进而从多个待预测行业中选择发展情况较好的行业推荐给用户进行投资，例如，用户考虑在教育、房地产和餐饮行业进行投资，通过对上述三个行业的排名结果进行预测，确定未来一段时间房地产行业的发展前景相对较好，则向用户推荐房地产行业，即建议用户在房地产行业进行投资。本申请主要应用于对投资用户进行行业推荐的场景。本申请实施例的执行主体为能够进行行业推荐的装置或设备，具体可以设置在客户端或者服务器一侧。
对于本申请实施例，当用户需要进行行业推荐时，可以在装置侧选择多个待预测行业，例如，用户想在房地产行业、教育行业和餐饮行业中选择发展前景相对较好的行业进行投资，则可以在装置中选择待预测行业为房地产行业、教育行业和餐饮行业，进而能够生成行业推荐请求，装置侧在接收到行业推荐请求后，会收集多个待预测行业在过去一段时间内的历史数据，并利用该历史数据对未来一段时间内的多个待预测行业的排名进行预测，具体在收集历史数据的过程中，为了提高行业排名的预测精度，可以收集待预测行业在不同维度下的历史数据，并根据预测的时间周期，对不同维度下的历史数据进行统计，以便根据统计后的不同维度下的历史数据对多个待预测行业的排名结果进行预测。
如分别收集房地产行业、教育行业和餐饮行业在2001-2005年内每周的订单量、生产总额、招聘人数和员工薪资，用上述数据对房地产行业、教育行业和餐饮行业在未来一周内的排名情况进行预测，具体地，根据2001-2005年内每周的订单量、生产总额、招聘人数和员工薪资和2001-2005年内包含的周数，计算2001-2005年内平均每周的订单量、生产总额、招聘人数和员工薪资，进而将计算的周平均订单量、周平均生成总额、周平均招聘人数和周平均员工薪资作为输入数据，用其预测未来一周内房地产行业、餐饮行业和教育行业之间的排名结果，以便根据该排名结果，确定发展前景较好的行业推荐给用户。
102、将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率。
其中，超额收益率预测模型具体可以为超额收益率回归预测模型，也可以为其他类型的预测模型，本申请实施例不做具体限定，对于本申请实施例，为了 挖掘历史数据中更多有效信息，能够从多个角度对行业排名结果进行预测，可以利于收集的历史数据，对未来一段时间内的超额收益率进行预测，进而从超额收益率这个角度来预测行业之间的排名。针对利于超额收益率归回预测模型进行收益率预测的具体过程，作为一种可选方式，步骤102具体包括：确定所述不同维度下的历史数据对应的权重参数；基于所述权重参数，将不同维度下的历史数据相加，得到多个所述待预测行业分别对应的超额收益率。
例如，将统计后的房地产行业的历史数据输入至预设超额收益率回归预测模型进行收益率预测，基于该预测模型中不同维度下的历史数据对应的权重值，将房地产行业的不同维度下的历史数据相加，得到房地产行业在未来一周内的超额收益率，同理可得到餐饮行业和教育行业在未来一周内的超额收益率，以便从超额收益率的角度对房地产行业、教育行业和餐饮行业的发展前景进行预测。
103、将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
其中，预设行业排名预测模型具体可以为学习排序模型，也可以为其他类型的预测模型，本申请实施例不做具体限定，针对对待预测行业进行排名的具体过程，作为一种可选实施方式，步骤103具体包括：对多个所述待预测行业进行两两分组，得到多组待预测行业；将所述多组待预测行业分别对应的超额收益率和历史数据共同输入至预设行业排名预测模型进行排名预测，得到所述多组待预测行业对应的组内排名结果；根据所述组内排名结果，确定多个所述待预测行业分别对应的行业排名结果。
例如，将房地产行业对应的超额收益率和历史数据作为输入数据(x 1，x 2，x 3，x 4，x 5)，其中，x 1代表房地产行业的周平均订单量，x 2代表房地产行业的周平均生成总额，x 3代表房地产房行业的周平均招聘人数，x 4代表房地产行业的周平均员工薪资，x 5代表房地产行业的超额收益率，同理可得到餐饮行业对应的输入数据为(y 1，y 2，y 3，y 4，y 5)和教育行业对应的输入数据(z 1，z 2，z 3，z 4，z 5)，之后将房地产行业、教育行业和餐饮行业进行两两分组得到，房地产行业和教育行业、教育行业和餐饮行业、房地产行业和餐饮行业，之后将每组待预测行业对应的超额收益率和历史数据共同输入至预设学习排序模型中进行排序预测，得到每组待预测行业对应的组内排序，进而根据该组内排序，确定房地产行业、教育行业和餐饮 行业之间的排序结果，具体预测过程见步骤204，由此不仅能够从历史数据的角度对行业排名进行预测，还能够从超额收益率的角度对行业排名进行预测，从而能够实现从多个角度对待预测行业的行业排名进行预测，确保行业排名结果的准确度，进一步地，根据该排序结果，能够确定待预测行业的业发展前景，以便向用户推荐发展前景较好的行业进行投资。
104、基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
对于本申请实施例，在确定行业排名结果之后，可以根据该行业排名结果从多个待预测行业中筛选排名符合预设条件要求的行业推荐给用户，例如，将多个待预测行业中排在首位的行业推荐给用户，或者将多个待预测行业中排名处于预设排名范围内的行业推荐用户，以便用户根据推荐的行业进行投资。
本申请实施例提供的一种行业推荐方法，与目前采用单一的预测模型对行业的发展前景进行预测，并根据预测结果为用户推荐相应行业的方式相比，本申请能够获取多个待预测行业的历史数据，并将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业对应的超额收益率，与此同时，将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，从而不仅能够从行业历史数据的角度对多个待预测行业的排名进行预测，还能够从超额收益率的角度对多个待预测行业的排名进行预测，进而能够根据预测的行业排名结果，确定多个待预测行业的发展前景，并基于该行业排名结果向用户推荐发展前景较好的行业，由此实现了从多个角度对行业的发展前景进行预测，提高了行业的推荐精度，能够为投资者提供更加全面、客观和有效的建议。
进一步的，为了更好的说明上述行业推荐的过程，作为对上述实施例的细化和扩展，本申请实施例提供了另一种行业推荐方法，如图2所示，所述方法包括：
201、获取多个待预测行业的历史数据。
对于本申请实施例，为了提高行业排名的预测精度，在对多个待预测行业的不同维度下的历史数据进行统计计算的过程中，不仅需要统计周平均订单量、周平均生产总额、周平均招聘人数和周平均员工薪资等，还需要统计周最大订单量、周最小订单量、订单量中位数、周最大生产总额、周最小生成总额、生成总额中位数、周最多招聘人数、周最少招聘人数、招聘人数中位数等，即在统计周 平均值的基础上，引入了周最大值、周最小和和中位数的概念，从而能够有利于挖掘历史数据中的更多有效信息。
此外，还可以统计月平均订单量、月最大订单量、月最小订单量、订单量中位数、月平均生产总额、月最大生产总额、月最小生成总额、生产总额中位数等，利用上述数据预测未来一个月内待预测行业的排名，需要说明的是，由于预测周期越短，预测的精度会越高，更有利于为投资者提供有效的建议，因此可以以周为单位预测待预测行业在未来连续几周为的排名，并基于该排名结果，能够确定待预测行业在未来连续几周内的发展情况，以便从长远给出投资者有效建议。
202、将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率。
其中，预设超额收益率预测模型具体可以为多元线性回归模型，将统计后的不同维度下的历史数据输入至多元线性回归模型进行收益率预测，基于多元线性回归模型中的权重参数，将统计后的不同维度下的历史数据相加，得到待预测行业在未来一段时间内的超额收益率，以便从超额收益率的角度对待预测行业的发展前景进行预测。
203、将所述历史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
对于本申请实施例，为了能够从多个角度对待预测行业的发展情况进行预测，还可以根据历史数据预测待预测行业在未来一段时间内超额收益率的变化趋势，进而从收益率变化趋势的角度预测行业的发展前景，基于此，步骤203具体包括：确定所述不同维度下的历史数据对应的行业排名名次；将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。其中，预设收益率变化趋势预测模型具体可以为逻辑回归模型，也可以为支持向量机、决策树等分类模型，收益率变化趋势包括收益趋势较好、收益趋势一般和收益趋势较差。
例如，首先确定房地产行业的周平均订单量、周平均生产总额、周平均招聘人数和周平均员工薪资在所有行业中的排名，同时确定教育行业和餐饮行业的周平均订单量、周平均生产总额、周平均招聘人数和周平均员工薪资在所有行业中的行名，进而将房地产行业、教育行业和餐饮行业的不同维度下的历史数据对应 的行业排名名次分别输入至预设逻辑回归模型中进行趋势预测，得到房地产行业、教育行业和餐饮行业属于不同收益率变化趋势的概率值，进而根据该概率值，分别确定房地产行业、教育行业和餐饮行业对应的收益率变化趋势。
在具体应用场景中，为了进一步提高收益率变化趋势的预测精度，需要检测出不同维度下的历史数据在过去某一段时间的异常值，进而通过对该异常值进行分析，能够更加准确地确定待预测行业对应的收益率变化趋势，基于此，在所述确定所述不同维度下的历史数据对应的行业排名名次之后，所述方法还包括：对所述不同维度下的历史数据进行异常检测，确定所述不同维度下的历史数据对应的异常检测结果，与此同时，所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：将所述异常检测结果和所述行业排名名次共同输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
例如，分别确定房地产行业在2001-2005年期间每周的订单量、生成总额、招聘人数和员工薪资，之后分别检测每周的订单量、生成总额、招聘人数和员工薪资是否在相应的预设范围内，如果超出预设范围，则确定其确定为异常值，并将该异常值作为输入数据输入至预设逻辑回归模型中进行趋势预测，如房地产行业对应的输入数据为(z 1，z 2，z 3，z 4，z 5，z 6，z 7，z 8)，其中，z 1，z 2，z 3，z 4，z 5分别代表房地产行业2001-2005年期间的周平均订单量、周平均生成总额、周平均招聘人数和周平均员工薪资，z 5，z 6，z 7，z 8分别代表订单量、生成总额、招聘人数和员工薪资对应的异常值，将其作为输入数据输入至预设回归模型中进行趋势预测，得到房地产行业对应的收益率变化趋势。
在具体应用场景中，为了能够进一步提高收益率变化趋势的预测精度，可以设定收益率变化趋势的预测过程由两个逻辑回归模型实现，即收益率变化趋势预测模型包括收益率正向变化趋势预测模型和收益率负向变化趋势预测模型，基于此，所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：将所述行业排名名次输入至预设收益率正向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率正向趋势预测结果；将所述行业排名名次输入至预设收益率负向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率负向趋势预测结果。
其中，预设收益率正向变化趋势预测模型可以为正向逻辑回归模型，正向逻辑回归模型对应的预测结果包括收益率趋势较好和收益率趋势一般，预设收益率负向变化趋势预测模型可以为负向逻辑回归模型，负向逻辑回归模型对应的预测结果包括收益率趋势一般和收益率趋势较差。对于本申请实施例，将不同维度下的历史数据对应的行业排名名次和异常值输入至预设正向逻辑回归模型进行趋势预测，得到待预测行业的收益率正向趋势预测结果，同时将不同维度下的历史数据对应的行业排名名次和异常值输入至预设负向逻辑回归模型进行趋势预测，得到待预测行业的收益率负向趋势预测结果，以便根据待预测行业对应的历史数据、超额收益率、收益率正向趋势预测结果和收益率负向趋势预测结果对待预测行业的发展前景进行预测。
204、将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
在具体应用场景中，步骤204具体包括：将所述收益率正向趋势预测结果、所述收益率负向趋势预测结果、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。其中，预设行业排名预测模型具体可以为RankNet排序模型，在利用RankNet排序模型对待预测行业的排名进行预测时，首先将待预测行业两两分成一组，得到多个行业组，之后分别将每组行业对应的收益率正向趋势预测结果、收益率负向趋势预测结果、超额收益率和历史数据输入至RankNet排序模型中进行预测，得到每组行业之间的排名结果，进而根据每组行业之间的排名结果，确定所有待预测行业之间的排名结果。
例如，待预测行业包括行业A、行业B和行业C，对待预测行业进行两两组合得到行业A和行业B、行业B和行业C、行业A和行业C，之后将每组行业对应的收益率正向趋势预测结果、收益率负向趋势预测结果、超额收益率和历史数据输入至RankNet排序模型中进行预测，根据待预测行业A是否比待预测行业B好，RankNet排序模型给出分类标签1或者0，即此时RankNet排序模型实际上为二分类器，能够输出待预测行业A比待预测行业B发展前景好的概率值，以及待预测行业B比待预测行业A发展前景好的概率值，进而依据该概率值，能够确定待预测行业A和待预测行业B之间的分类标签，根据该分类标签，确定待预测 行业A和待遇测行业B之间的排序，如输出的分类标签为1，确定待预测行业A排列在待预测行业B之前，同理能够确定行业B和行业C之间的排列顺序，以及行业A和行业C之间的排列顺序，进而根据得到的每组行业对应的排列顺序，能够确定行业A、行业B和行业C之间的排列顺序，即能够确定行业A、行业B和行业C对应的行业排名。
进一步地，针对RankNet排序模型的训练过程，可以预先收集样本行业对应的收益率正向趋势预测结果、收益率负向趋势预测结果、超额收益率和历史数据，并将样本行业两两划分成一组，根据样本行业在过去一段时间内的发展情况，对每组样本行业进行标注，将标注后的样本行业作为训练集，通过对该训练集进行训练，构建RankNet排序模型，训练过程中的具体公式如下：
Figure PCTCN2021097272-appb-000001
其中，待预测行业集合S中的第i个行业记做U i，其对应的特征向量记做x i，对于一个行业组U i，U j，RankNet排序模型通过一个全连接层将输入的行业特征向量映射到f(x)，得到S i＝f(x i)，S j＝f(x j)，并将U i比U j发展前景好的概率记做P i,j，进一步地，构建度量损失函数，如下：
Figure PCTCN2021097272-appb-000002
其中，C i,j为度量损失函数，
Figure PCTCN2021097272-appb-000003
代表真是概率值，记做：
Figure PCTCN2021097272-appb-000004
Figure PCTCN2021097272-appb-000005
其中，S i,j＝{+1,0}代表行业组对应的标签，即行业U i的发展前景是否好于行业U j的发展前景，最终可得到度量损失函数，进而能够利用该度量损失函数对参数σ进行优化，构建RankNet排序模型。
205、基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
对于本申请实施例，可以将多个待预测行业中排名第一的行业推荐给用户进行投资，或者将排名处于预设范围内的行业推荐给用户进行参考，如共存在10个待预测行业，将排名前3的行业推荐给用户进行参考。
本申请实施例提供的另一种行业推荐方法，与目前采用单一的预测模型对行业的发展前景进行预测，并根据预测结果为用户推荐相应行业的方式相比，本 申请能够获取多个待预测行业的历史数据，并将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率，与此同时，将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，从而不仅能够从行业历史数据的角度对多个待预测行业的排名进行预测，还能够从超额收益率的角度对多个待预测行业的排名进行预测，进而能够根据预测的行业排名结果，确定多个待预测行业的发展前景，并基于该行业排名结果向用户推荐发展前景较好的行业，由此实现了从多个角度对行业的发展前景进行预测，提高了行业的推荐精度，能够为投资者提供更加全面、客观和有效的建议。
进一步地，作为图1的具体实现，本申请实施例提供了一种行业推荐装置，如图3所示，所述装置包括：获取单元31、第一预测单元32、第二预测单元33和推荐单元34。
所述获取单元31，设置为获取多个待预测行业的历史数据。所述获取单元31是本装置中获取多个待预测行业的历史数据的主要功能模块。
所述第一预测单元32，设置为将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率。所述第一预测单元32是本装置中将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率的主要功能模块，也是核心模块。
所述第二预测单元33，设置为将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。所述第二预测单元33是本装置中将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果的主要功能模块，也是核心模块。
所述推荐单元34，设置为基于所述行业排名结果从多个所述待预测行业中推荐相应行业。所述推荐单元34是本装置中基于所述行业排名结果从多个所述待预测行业中推荐相应行业的主要功能模块。
在具体应用场景中，为了提高对待预测行业的预测精度，所述第一预测单元32，还用于将所述历史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
所述第二预测单元33，具体设置为将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
在具体应用场景中，所述历史数据为不同维度下的历史数据，如图4所示，所述第一预测单元32，包括：确定模块321和预测模块322。
所述确定模块321，设置为确定所述不同维度下的历史数据对应的行业排名名次。
所述预测模块322，设置为将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
在具体应用场景中，为了提高收益率变化趋势的预测精度，所述第一预测单元32，还包括：检测模块323。
所述检测模块323，设置为对所述不同维度下的历史数据进行异常检测，确定所述不同维度下的历史数据对应的异常检测结果。
所述预测模块322，具体设置为将所述异常检测结果和所述行业排名名次共同输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
在具体应用场景中，所述收益率变化趋势预测模型包括收益率正向变化趋势预测模型和收益率负向变化趋势预测模型，所述预测模块322，还设置为将所述行业排名名次输入至预设收益率正向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率正向趋势预测结果。
所述预测模块322，还设置为将所述行业排名名次输入至预设收益率负向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率负向趋势预测结果。
所述第二预测单元33，具体设置为将所述收益率正向趋势预测结果、所述收益率负向趋势预测结果、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
在具体应用场景中，为了确定待预测行业对应的行业排名结果，所述第二预测单元33，包括：分组模块331、预测模块332和确定模块333。
所述分组模块331，设置为对多个所述待预测行业进行两两分组，得到多组待预测行业。
所述预测模块332，设置为将所述多组待预测行业分别对应的超额收益率和历史数据共同输入至预设行业排名预测模型进行排名预测，得到所述多组待预测行业对应的组内排名结果。
所述确定模块333，设置为根据所述组内排名结果，确定多个所述待预测行业分别对应的行业排名结果。
在具体应用场景中，所述第一预测单元32，还包括相加模块324。
所述确定模块321，还设置为确定所述不同维度下的历史数据对应的权重参数。
所述相加模块324，设置为基于所述权重参数，将不同维度下的历史数据相加，得到多个所述待预测行业分别对应的超额收益率。
需要说明的是，本申请实施例提供的一种行业推荐装置所涉及各功能模块的其他相应描述，可以参考图1所示方法的对应描述，在此不再赘述。
基于上述如图1所示方法，相应的，本申请实施例还提供了一种计算机可读存储介质，其上存储有计算机可读指令，该程序被处理器执行时实现以下步骤：获取多个待预测行业的历史数据；将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及基于所述行业排名结果从多个所述待预测行业中推荐相应行业。其中，所述计算机可读存储介质可以是非易失性，也可以是易失性。
基于上述如图1所示方法和如图3所示装置的实施例，本申请实施例还提供了一种计算机设备的实体结构图，如图5所示，该计算机设备包括：处理器41、存储器42、及存储在存储器42上并可在处理器上运行的计算机可读指令，其中存储器42和处理器41均设置在总线43上所述处理器41执行所述程序时实现以下步骤：获取多个待预测行业的历史数据；将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
通过本申请的技术方案，本申请能够获取多个待预测行业的历史数据，并将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率，与此同时，将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，从而不仅能够从行业历史数据的角度对多个待预测行业的排名进行预测，还能够从超额收益率的角度对多个待预测行业的排名进行预测，进而能够根据预测的行业排名结果，确定多个待预测行业的发展前景，并基于该行业排名结果向用户推荐发展前景较好的行业，由此实现了从多个角度对行业的发展前景进行预测，提高了行业的推荐精度，能够为投资者提供更加全面、客观和有效的建议。
显然，本领域的技术人员应该明白，上述的本申请的各模块或各步骤可以用通用的计算装置来实现，它们可以集中在单个的计算装置上，或者分布在多个计算装置所组成的网络上，可选地，它们可以用计算装置可执行的程序代码来实现，从而，可以将它们存储在存储装置中由计算装置来执行，并且在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤，或者将它们分别制作成各个集成电路模块，或者将它们中的多个模块或步骤制作成单个集成电路模块来实现。这样，本申请不限制于任何特定的硬件和软件结合。
以上所述仅为本申请的优选实施例而已，并不用于限制本申请，对于本领域的技术人员来说，本申请可以有各种更改和变化。凡在本申请的精神和原则之内，所作的任何修改、等同替换、改进等，均应包括在本申请的保护范围之内。

Claims (20)

  1. 一种行业推荐方法，其中，包括：
    获取多个待预测行业的历史数据；
    将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
    将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
    基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
  2. 根据权利要求1所述的方法，其中，在将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率之后，所述方法还包括：
    将所述历史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势；
    所述将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
  3. 根据权利要求2所述的方法，其中，所述历史数据为不同维度下的历史数据，所述将所述历史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    确定所述不同维度下的历史数据对应的行业排名名次；
    将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
  4. 根据权利要求3所述的方法，其中，在所述确定所述不同维度下的历史数据对应的行业排名名次之后，所述方法还包括：
    对所述不同维度下的历史数据进行异常检测，确定所述不同维度下的历史数据对应的异常检测结果；
    所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    将所述异常检测结果和所述行业排名名次共同输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
  5. 根据权利要求3所述的方法，其中，所述收益率变化趋势预测模型包括收益率正向变化趋势预测模型和收益率负向变化趋势预测模型，所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    将所述行业排名名次输入至预设收益率正向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率正向趋势预测结果；
    将所述行业排名名次输入至预设收益率负向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率负向趋势预测结果；
    所述将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    将所述收益率正向趋势预测结果、所述收益率负向趋势预测结果、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
  6. 根据权利要求1所述的方法，其中，所述将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    对多个所述待预测行业进行两两分组，得到多组待预测行业；
    将所述多组待预测行业分别对应的超额收益率和历史数据共同输入至预设行业排名预测模型进行排名预测，得到所述多组待预测行业对应的组内排名结果；
    根据所述组内排名结果，确定多个所述待预测行业分别对应的行业排名结果。
  7. 根据权利要求3所述的方法，其中，所述将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率，包括：
    确定所述不同维度下的历史数据对应的权重参数；
    基于所述权重参数，将不同维度下的历史数据相加，得到多个所述待预测行业分别对应的超额收益率。
  8. 一种行业推荐装置，其中，包括：
    获取单元，设置为获取多个待预测行业的历史数据；
    第一预测单元，设置为将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
    第二预测单元，设置为将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
    推荐单元，设置为基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
  9. 一种计算机可读存储介质，其上存储有计算机可读指令，其中，所述计算机可读指令被处理器执行时实现行业推荐方法，包括：
    获取多个待预测行业的历史数据；
    将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
    将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
    基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
  10. 根据权利要求9所述的计算机可读存储介质，其中，所述计算机可读指令被处理器执行时实现在将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率之后，所述方法还包括：
    将所述历史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势；
    所述将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
  11. 根据权利要求10所述的计算机可读存储介质，其中，所述计算机可读指令被处理器执行时实现所述历史数据为不同维度下的历史数据，所述将所述历 史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    确定所述不同维度下的历史数据对应的行业排名名次；
    将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
  12. 根据权利要求11所述的计算机可读存储介质，其中，所述计算机可读指令被处理器执行时实现在所述确定所述不同维度下的历史数据对应的行业排名名次之后，所述方法还包括：
    对所述不同维度下的历史数据进行异常检测，确定所述不同维度下的历史数据对应的异常检测结果；
    所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    将所述异常检测结果和所述行业排名名次共同输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
  13. 根据权利要求11所述的计算机可读存储介质，其中，所述计算机可读指令被处理器执行时实现所述收益率变化趋势预测模型包括收益率正向变化趋势预测模型和收益率负向变化趋势预测模型，所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    将所述行业排名名次输入至预设收益率正向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率正向趋势预测结果；
    将所述行业排名名次输入至预设收益率负向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率负向趋势预测结果；
    所述将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    将所述收益率正向趋势预测结果、所述收益率负向趋势预测结果、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
  14. 根据权利要求9所述的计算机可读存储介质，其中，所述计算机可读 指令被处理器执行时实现所述将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    对多个所述待预测行业进行两两分组，得到多组待预测行业；
    将所述多组待预测行业分别对应的超额收益率和历史数据共同输入至预设行业排名预测模型进行排名预测，得到所述多组待预测行业对应的组内排名结果；
    根据所述组内排名结果，确定多个所述待预测行业分别对应的行业排名结果。
  15. 一种计算机设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机可读指令，其中，所述计算机可读指令被处理器执行时实现行业推荐方法，包括：
    获取多个待预测行业的历史数据；
    将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率；
    将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果；以及
    基于所述行业排名结果从多个所述待预测行业中推荐相应行业。
  16. 根据权利要求15所述的计算机设备，其中，所述计算机可读指令被处理器执行时实现在将所述历史数据输入至预设超额收益率预测模型进行收益率预测，得到多个所述待预测行业分别对应的超额收益率之后，所述方法还包括：
    将所述历史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势；
    所述将所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
  17. 根据权利要求16所述的计算机设备，其中，所述所述计算机可读指令被处理器执行时实现所述历史数据为不同维度下的历史数据，所述将所述历史数据输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业 分别对应的收益率变化趋势，包括：
    确定所述不同维度下的历史数据对应的行业排名名次；
    将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
  18. 根据权利要求17所述的计算机设备，其中，所述所述计算机可读指令被处理器执行时实现在所述确定所述不同维度下的历史数据对应的行业排名名次之后，所述方法还包括：
    对所述不同维度下的历史数据进行异常检测，确定所述不同维度下的历史数据对应的异常检测结果；
    所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    将所述异常检测结果和所述行业排名名次共同输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势。
  19. 根据权利要求17所述的计算机设备，其中，所述所述计算机可读指令被处理器执行时实现所述收益率变化趋势预测模型包括收益率正向变化趋势预测模型和收益率负向变化趋势预测模型，所述将所述行业排名名次输入至预设收益率变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率变化趋势，包括：
    将所述行业排名名次输入至预设收益率正向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率正向趋势预测结果；
    将所述行业排名名次输入至预设收益率负向变化趋势预测模型进行趋势预测，得到多个所述待预测行业分别对应的收益率负向趋势预测结果；
    所述将所述收益率变化趋势、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    将所述收益率正向趋势预测结果、所述收益率负向趋势预测结果、所述超额收益率和所述历史数据共同输入至预设行业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果。
  20. 根据权利要求15所述的计算机设备，其中，所述所述计算机可读指令被处理器执行时实现所述将所述超额收益率和所述历史数据共同输入至预设行 业排名预测模型进行排名预测，得到多个所述待预测行业分别对应的行业排名结果，包括：
    对多个所述待预测行业进行两两分组，得到多组待预测行业；
    将所述多组待预测行业分别对应的超额收益率和历史数据共同输入至预设行业排名预测模型进行排名预测，得到所述多组待预测行业对应的组内排名结果；根据所述组内排名结果，确定多个所述待预测行业分别对应的行业排名结果。
PCT/CN2021/097272 2021-04-25 2021-05-31 行业推荐方法、装置、计算机设备及存储介质 WO2022227213A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110447074.XA CN113159964A (zh) 2021-04-25 2021-04-25 行业推荐方法、装置、计算机设备及存储介质
CN202110447074.X 2021-04-25

Publications (1)

Publication Number Publication Date
WO2022227213A1 true WO2022227213A1 (zh) 2022-11-03

Family

ID=76870315

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/097272 WO2022227213A1 (zh) 2021-04-25 2021-05-31 行业推荐方法、装置、计算机设备及存储介质

Country Status (2)

Country Link
CN (1) CN113159964A (zh)
WO (1) WO2022227213A1 (zh)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107909433A (zh) * 2017-11-14 2018-04-13 重庆邮电大学 一种基于大数据移动电子商务的商品推荐方法
CN109800928A (zh) * 2019-03-25 2019-05-24 广州大学 一种基于会计信息相关性的投资组合预测方法及系统
CN111738856A (zh) * 2020-06-24 2020-10-02 四川长虹电器股份有限公司 一种股票舆情投资决策分析方法及装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107909433A (zh) * 2017-11-14 2018-04-13 重庆邮电大学 一种基于大数据移动电子商务的商品推荐方法
CN109800928A (zh) * 2019-03-25 2019-05-24 广州大学 一种基于会计信息相关性的投资组合预测方法及系统
CN111738856A (zh) * 2020-06-24 2020-10-02 四川长虹电器股份有限公司 一种股票舆情投资决策分析方法及装置

Also Published As

Publication number Publication date
CN113159964A (zh) 2021-07-23

Similar Documents

Publication Publication Date Title
US20200401939A1 (en) Systems and methods for preparing data for use by machine learning algorithms
WO2019165673A1 (zh) 一种报销单风险预测方法、装置、终端设备及存储介质
Chai et al. Application of decision-making techniques in supplier selection: A systematic review of literature
CN111324642A (zh) 一种面向电网大数据分析的模型算法选型与评价方法
CN107766929B (zh) 模型分析方法及装置
Malhotra et al. Software effort prediction using statistical and machine learning methods
CN110956273A (zh) 融合多种机器学习模型的征信评分方法及系统
EP4020315A1 (en) Method, apparatus and system for determining label
CN113537807B (zh) 一种企业智慧风控方法及设备
WO2017071369A1 (zh) 一种预测用户离网的方法和设备
Chaudhary et al. Comparative Analysis of Entropy Weight Method and C5 Classifier for Predicting Employee Churn
Jain et al. A data mining solution to predict campus placement
Bi et al. Accurate Prediction of Workloads and Resources with Multi-head Attention and Hybrid LSTM for Cloud Data Centers
CN113743615A (zh) 简化机器学习的特征移除框架
CN112836750A (zh) 一种系统资源分配方法、装置及设备
WO2022227213A1 (zh) 行业推荐方法、装置、计算机设备及存储介质
CN116523301A (zh) 基于电商大数据进行风险评级预测的系统
US11782923B2 (en) Optimizing breakeven points for enhancing system performance
Malleswari et al. Enhanced SS-FIM Algorithm For High Utility Uncertain Itemsets
WO1992017853A2 (en) Direct data base analysis, forecasting and diagnosis method
CN111126629A (zh) 模型的生成方法、刷单行为识别方法、系统、设备和介质
Cui et al. Answer aggregation for crowdsourcing microtasks using approximate global optimal searching
US10699229B2 (en) User segmentation for cost-based deployment of enterprise infrastructure
Huang et al. A clustering-based method for business hall efficiency analysis
CN117787569B (zh) 一种智能辅助评标方法及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21938683

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21938683

Country of ref document: EP

Kind code of ref document: A1