CN108197729A - Value investment analysis method, equipment and storage medium based on machine learning - Google Patents
Value investment analysis method, equipment and storage medium based on machine learning Download PDFInfo
- Publication number
- CN108197729A CN108197729A CN201711402200.XA CN201711402200A CN108197729A CN 108197729 A CN108197729 A CN 108197729A CN 201711402200 A CN201711402200 A CN 201711402200A CN 108197729 A CN108197729 A CN 108197729A
- Authority
- CN
- China
- Prior art keywords
- target
- plate
- similarity
- value
- market
- 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
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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Technology Law (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Tourism & Hospitality (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of Value investment analysis method, equipment and storage medium based on machine learning, the described method comprises the following steps：Analysis target is subjected to Plate division, calculates analysis target and the similarity of the plate；Structure builds neural network model based on neural network theory, and the neural network module includes at least input layer, hidden layer and output layer, and the input layer is the linear input layer of multidimensional；Market of the analysis target lower period in each plate as input layer parameter, are predicted using the financial factor in a period and plate similarity in the analysis target；Similarity-Weighted is carried out to each plate forecast for market tendency value to be averaged, and takes Similarity-Weighted average value as the analysis target forecast for market tendency value.The present invention analyzes target synthesis market by building multilayer neural network, to analyze the similarity of target history financial data and plate to predict described in next period, make moderate progress to the Stability and veracity of forecast for market tendency long-term in analysis target.
Description
Technical field
The present invention relates to financial field, particularly a kind of Value investment analysis method, equipment based on machine learning
And storage medium.
Background technology
Prediction is a kind of important data analysis form, and data prediction is one and is made of study stage and forecast period
Two benches process, in the first stage i.e. the study stage by from training set " study " or analysis sorting algorithm structural classification device.
The tuple (sample) of data and class symbol associated there constitute training set in database.Due to the presence of class label,
Entire training process has supervision, can be regarded as one mapping function of study the study stage, input and output are all set.
And in second stage, mainly test set data are predicted using model, the indexs such as assessment prediction accuracy rate.It is so-called accurate
Rate refers to the percentage of whole tuples shared by the correct test set tuple of grader classification.
Most scholars think, stock index is for the relevant data of time series, i.e., by research history data, to stock
The development of ticket index can obtain certain prediction result, for finance, due to the high yield and high risk of stock market
And the characteristic deposited, requirement of the Prediction of Stock Index for algorithm is very high, and conventional method mainly includes Fundamental Analysis and technology analytic approach,
Shares changing tendency is analyzed by the market factor such as supply-demand relationship and statistical analysis respectively, prediction difficulty is larger, and prediction result
Accuracy is not high.
With the development of internet industry, information technology is occupied an leading position, and securities market is towards the market direction of modernization
Development.Present Shanghai and Shenzhen listed company alreadys exceed thousands of families, however the income of equity investment to risk be often it is directly proportional, i.e.,
Investment return is higher, may risk risk is bigger.Therefore, the research of Stock Market Forecasting method have extremely important application value and
Theory significance.Always there are many conventional analytical techniques, it should say these traditional technology analysis methods on stock analysis also
To achieve larger achievement, however, it is not difficult to find that these existing theoretical and methods be also there is it is very big the defects of
, they be invariably using static method, qualitative description it is more, quantitative description lacks, and many factors for influencing stock market are cut
It splits and carrys out single analysis.Therefore, these limitations cause these methods cannot be effective in Stock Price Fluctuation changeable
Accurately hold the variation of stock price.Therefore need to explore the complexity and regularity of stock market fluctuations, and according to it
Regularity design is a series of easy to operate, the enough high forecasting softwares of precision, and investors avoid risk to be vast.
Invention content
To overcome drawbacks described above, the present invention provides a kind of Value investment analysis method based on machine learning, equipment and deposits
Storage media, by the future prospects for building Neural Network Prediction target.
For this purpose, the wherein embodiment of the present invention provides a kind of Value investment analysis method based on machine learning, it is described
Method is suitable for performing in computing device, includes the following steps：
(1) analysis target is subjected to Plate division, calculates analysis target and the similarity of the plate；
(2) structure builds neural network model based on neural network theory, and the neural network module includes at least input
Layer, hidden layer and output layer, the input layer are the linear input layer of multidimensional；
(3) using the financial factor in a period and plate similarity in the analysis target, as input layer parameter, prediction should
Analyze market of the target lower period in each plate；
(4) it carries out Similarity-Weighted to each plate forecast for market tendency value to be averaged, takes Similarity-Weighted average value conduct
The forecast for market tendency value of the analysis target；
(5) result is shown.
In the step (1), an analysis target can be divided into multiple plates, the analysis target division method of slab packet
It includes：
The analysis basic side information of target and technological side data information are obtained, extracts basic side information and technology face data letter
Keyword in breath；
Merge the word that synonym or similarity in the keyword meet threshold value, generate plate characteristic word, the plate
Feature Words are the plate belonging to the analysis target.
Further, in the step (1), the computational methods of the analysis target and the similarity of affiliated plate include：
Extraction and analysis target basic side information and technological side data information keyword；
Calculate the similarity of the keyword and affiliated plate characteristic word.
In the step (3), the financial factor includes the financial factor and includes earnings per share, net profit on sales
Rate, net assets income ratio, per share management road net profit, operating income, main management industry year-on-year growth rate, prime operating revenue ring ratio
Growth rate, operating profit, net profit, total profit, net profit year-on-year growth rate, net profit sequential growth rate, cash flow, money
Produce debt ratio in one or more combination or it is multinomial between arithmetic generate the factor；
The market include opening price, closing price, highest price, lowest price, exchange hand, transaction value, turnover rate, circulation city
One or more combinations in value, total market capitalisation, shareholder's number or it is multinomial between the market factor that generates of arithmetic.
Further, it in the step (3), before the financial factor parameter of input, further includes：The financial factor is clustered
The step of, the cluster generation debt paying ability factor, the operation ability factor, the profitability factor, enterprise development capability facfor.
Further, in the step (4), the forecast for market tendency value of the analysis target：
Wherein, N is the plate quantity belonging to the analysis target,J-th of prediction of the plate market for the i-th plate
Value, xiFor the analysis target and the similarity of affiliated plate.
Further, a kind of Value investment analysis method based on machine learning of the present invention is further comprising the steps of：
It obtains with reference to target, by analysis target with establishing interaction relation with reference to target；
It calculates and more described with reference to target and the financial similarity for analyzing target, if the financial similarity of the two is poor
Different is more than threshold value, then provides the warning prompt information of analysis target, wherein, the difference includes difference, ratio or relative value.
Further, the screening technique with reference to target includes：
Selection is with reference to the range of target：Including all targets in the plate with analyzing target similarity maximum as the first mark
's；
It obtains with reference to target, calculate the first target and analyzes the plate similarity of target, it is more than threshold value or phase to take similarity
It is used as like the first maximum target of degree with reference to target, wherein, the plate similarity is by extracting first target and analysis
The plate keyword of target and the similarity acquisition for calculating the keyword.
Further, the financial similarity is by calculating the financial factor with reference to target and the analysis target
Similarity obtain, the similarity calculating method include standardization Euclidean distance, beat up it is several in distance, manhatton distance, skin
Ademilson correlation.
Another embodiment of the present invention provides a kind of computing device, including：
One or more processors；
Memory；And
One or more programs, wherein one or more of programs are stored in the memory and are configured as by institute
One or more processors execution is stated, one or more of programs include performing the Value investment based on machine learning point
Either method in analysis method.
Another embodiment of the present invention provides a kind of computer storage media, and the storage medium is stored with one or more
Program, one or more of programs include instruction, and described instruction is when executed by a computing apparatus so that the computing device is held
Either method in Value investment analysis method of the row based on machine learning.
Compared with prior art, the present invention is based on the Value investment analysis method of machine learning, equipment and storage medium tools
It has the advantage that：
1. the present invention is by building multilayer neural network, to analyze the history financial data of target and target and the phase of plate
Like degree as the linear input layer of multidimensional, market of the target in each plate are analyzed described in next period, and to described each to predict
Plate forecast for market tendency value carries out Similarity-Weighted and is averaged, and takes forecast for market tendency of the Similarity-Weighted average value as the analysis target
Value, is combined into multidimensional with history financial data and plate similarity and linearly inputs layer building network model, to long in analysis target
The Stability and veracity of the forecast for market tendency of phase makes moderate progress, and can more accurately and rapidly be predicted；
2. the target of the analysis target and similarity maximum is established into interaction relation, if the difference of the forecast for market tendency value of the two
Different is more than threshold value, then warning prompt information is provided, to be supplied to the analysis target that user's early warning is investigated may be in financial index
On have a greater change or it is possible that the suspicion played tricks.
Description of the drawings
Fig. 1 is Value investment analysis method flow chart of the embodiment of the present invention one based on machine learning.
Fig. 2 is Value investment analysis method flow chart of the embodiment of the present invention two based on machine learning.
Specific embodiment
The present invention offer it is a kind of suitable for performed in computing device the Value investment analysis method based on machine learning,
Equipment and storage medium, the present invention is by building multilayer neural network, to analyze the history financial data of target and the plate of target
Block similarity analyzes market of the target in each plate, and to described as the linear input layer of multidimensional to predict described in next period
Each plate forecast for market tendency value carries out Similarity-Weighted and is averaged, and takes market of the Similarity-Weighted average value as the analysis target
Predicted value is combined into multidimensional with history financial data and plate similarity and linearly inputs layer building network model, to analyzing target
The Stability and veracity of medium-term and long-term forecast for market tendency makes moderate progress, and can more accurately and rapidly be predicted.
Value investment analysis method the present invention is based on machine learning is suitable for the forecast for market tendency of stock, below in conjunction with this
Attached drawing in inventive embodiments is clearly and completely described the technical solution in the embodiment of the present invention, it is clear that described
Embodiment be only part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ability
Domain those of ordinary skill all other embodiments obtained under the premise of creative work is not made, belong to the present invention
The range of protection.
Embodiment one
The present invention provides a kind of Value investment analysis method based on machine learning, suitable for being performed in computing device.Ginseng
Fig. 1 is examined, the Value investment analysis method flow chart the present invention is based on machine learning is shown, includes the following steps：
Step 101, analysis target is subjected to Plate division, calculates analysis target and the similarity of the plate.
Specifically, an analysis target can be divided into multiple and different plates simultaneously, be marked in the present embodiment by obtaining analysis
Basic side information and technological side data information, extract the keyword in basic side information and technological side data information, merge
Synonym or similarity in keyword meet the word of threshold value, generate plate characteristic word, and find from plate keywords database with
For word similar in plate characteristic word as plate title, the plate title is the plate belonging to the analysis target.
By taking domestic A share market as an example, in the present embodiment, analyze target stock information can from great wisdom, it is sensible letter,
The platforms such as straight flush obtain, and corresponding information includes the basic side information of stock and technological side data information, basic side letter
Breath includes but not limited to the title of the affiliated company of stock, stock code, main business, industry, product, financial index, newest disappears
Breath, shareholder, concept subject matter.Stock technological side data information, the including but not limited to price of stock, amount of increase and amount of decrease, change closing time
Hand rate, exchange hand, transaction value, amount ratio.The period of above-mentioned basic side information and technological side information crosses can be by user by happiness
Good setting can be the last week, preceding January, the previous year.Information time zone is shorter, and Plate division can be more accurate, to user's
Reference significance bigger.In the present embodiment, the market data for rejecting suspension personal share are further included.
Synonym in above-mentioned keyword or near synonym are merged, generate plate characteristic word.Specifically, meter can be passed through
The similarity between keyword is calculated to merge, by setting similarity threshold, plate is merged into the word for meeting threshold value
Block feature word, and found in the plate keyword stored in dictionary with word similar in plate characteristic word as plate title,
The plate belonging to analysis target is referred to as with the plate name in specific embodiment.
Meanwhile the similarity of each plate characteristic word and affiliated plate in stock is calculated, using similarity as subsequent training
Data.
Step 102, neural network model is built based on neural network theory, the neural network module includes at least input
Layer, hidden layer and output layer.
Using deep learning algorithm, construction multilayer neural network model is for machine learning, the multilayer neural network model
Including at least input layer, hidden layer, output layer.In the present embodiment, using the linear input layer of various dimensions as input layer, with
Hidden layer of the Sigmoid functions as activation primitive, the output layer of the Tanh functions as activation primitive.
Specifically, which is the linear input layer of multidimensional, to analyze the financial factor of target history cycle and plate phase
Like degree as input layer parameter, market of the analysis target lower period in each plate are predicted.In the present embodiment, the period can be with
It is finance cycle, that is, refers to that a financial statement is published to the time of next financial statement publication, specifically may be set to season.
The financial factor is the financial data of listed company's season quarterly report, and the financial statement issued by inquiring company obtains, and can also lead to
Cross third-party platform quotient, such as the F10 pages of straight flush, sensible letter, great wisdom.
Using the financial factor in quarterly report and plate similarity as the input parameter of neural network model input layer, the finance
The factor includes earnings per share, net profit on sales rate, net assets income ratio, per share management road net profit, operating income, main management
It is industry year-on-year growth rate, prime operating revenue sequential growth rate, operating profit, net profit, total profit, net profit year-on-year growth rate, net
In profit sequential growth rate, cash flow, asset-liability ratio one or more combination or it is multinomial between arithmetic
The factor of generation.
Preferably, it before the financial factor parameter of input, further includes：The step of financial factor is clustered, cluster generation
The debt paying ability factor, the operation ability factor, the profitability factor, enterprise development capability facfor, with one in the above-mentioned factor or
Multinomial combination is used as input parameter.
It is corresponding by the one or more financial factors and analysis target that input multiple history seasons in specific embodiment
Plate similarity, using the data of above-mentioned various dimensions as training data, the market of output include opening price, closing price, highest
One or more combinations in valency, lowest price, exchange hand, transaction value, turnover rate, circulation value, total market capitalisation, shareholder's number or
The market factor that arithmetic between person is multinomial generates.Similarity-Weighted is carried out to each plate forecast for market tendency value to put down
, forecast for market tendency value of the Similarity-Weighted average value as the analysis target is taken, with reference to practical analysis target market adjustment
And build neural network model.
Preferably, (i.e. the 5th season lower first quarter will be predicted with the financial factor of first four season quarterly report and plate similarity
Degree) market, choose net profit, operating income, business growth rate, profit growth rate, asset-liability ratio, plate similarity conduct
It is corresponding to test output analysis target as input layer by training for the linear input layer that the input parameter of model, i.e. dimension are 4*6
Next period forecasting market data list of plate.Similarity-Weighted is carried out to each plate forecast for market tendency value to be averaged, and is taken
Forecast for market tendency value of the Similarity-Weighted average value as the analysis target, adjusts and builds with reference to practical analysis target market
Neural network model.
Specifically, the market data of plate quantity generation respective numbers divided by analysis target, the plate quantity of division
More, the synthesis market of prediction are more accurate.
Step 103, to analyze the financial factor in a period and plate similarity in target, as input layer parameter, prediction should
Analyze market of the target lower period in each plate.
Using analyze target last quarter the financial factor and plate similarity as neural network model input layer input
Parameter, to predict the market of the lower first quarter.The financial factor of the input includes earnings per share, net profit on sales rate, Net Gains of Asset
Rate, per share management road net profit, operating income, main management industry year-on-year growth rate, prime operating revenue sequential growth rate, business profit
In profit, net profit, total profit, net profit year-on-year growth rate, net profit sequential growth rate, cash flow, asset-liability ratio
One or more combine or it is multinomial between arithmetic generate the factor.
Preferably, the net profit of selection quarterly report last quarter, operating income, business growth rate, profit growth rate, assets are born
The input parameter of debt rate, plate similarity as model exports the data list of the forecast for market tendency value of each plate.The analysis mark
Each plate forecast for market tendency value for the analysis target opening price of the lower first quarter, closing price, highest price, most in the plate
At a low price, one or more combinations or multinomial in exchange hand, transaction value, turnover rate, circulation value, total market capitalisation, shareholder's number
Between arithmetic generate the market factor preset value.
Step 104, it carries out Similarity-Weighted to each plate forecast for market tendency value to be averaged, takes Similarity-Weighted average value
Forecast for market tendency value as the analysis target.
Similarity-Weighted is carried out to each plate forecast for market tendency value to be averaged, and is taken described in the conduct of Similarity-Weighted average value
The synthesis forecast for market tendency value of target is analyzed, the synthesis forecast for market tendency value of the analysis target includes the opening price of the lower first quarter, receives
One or more in disk valency, highest price, lowest price, exchange hand, transaction value, turnover rate, circulation value, total market capitalisation, shareholder's number
Combination or it is multinomial between arithmetic generate the market factor.
In specific embodiment, preset analysis target is A-share ticket, which can be according to basic side information and technological side number
It is believed that breath is divided into N number of plate, respectively N1, N2……Ni, the similarity of A-share ticket and above-mentioned i-th of plate is xi, while profit
The financial factor of last quarter and plate similarity x are inputted with the neural network model of structureiPredict A-share ticket in i-th of plate
In the lower first quarter forecast for market tendency valueThe market include opening price, closing price, highest price, lowest price, exchange hand, transaction value,
One or more combinations in turnover rate, circulation value, total market capitalisation, shareholder's number or it is multinomial between arithmetic generate
The market factor predicted value.Similarity-Weighted is carried out to the forecast for market tendency value of above-mentioned each plate, takes Similarity-Weighted average value
Synthesis forecast for market tendency value P as the analysis target lower first quarterj.Synthesis forecast for market tendency value PjCalculation formula it is as follows：
In specific embodiment,With PjRepresent corresponding market,For analysis target the i-th plate plate market
J-th of predicted value,Including opening price, closing price, highest price, lowest price, exchange hand, transaction value, turnover rate, circulation value,
One or more combinations in total market capitalisation, shareholder's number or it is multinomial between the prediction of the market factor that generates of arithmetic
It is worth, then corresponding PjThe integrated forecasting value of corresponding market for the analysis target.
Specifically, A-share ticket can be divided into N1, 3 plates of N2, N3, correspondence and the similarity of above three plate are respectively
0.6,0.7,0.8, it is 3.2 yuan using opening price predicted value of the Neural Network model predictive A-share ticket in plate N1, in plate N2
In opening price predicted value for 5.2 yuan, the opening price predicted value in plate N1 is 4.5 yuan, then the lower first quarter of A-share ticket is comprehensive
Closing opening price forecast for market tendency value is：
P=(0.6*3.2+0.7*5.2+0.8*4.5)/(0.6+0.7+0.8)=4.36 yuan
Step 105, result is shown.
Show analysis result, specifically, include display analysis each plate of target forecast for market tendency value, weighted value and
Comprehensive forecast for market tendency value, above-mentioned display mode include single On the Tape or superposition On the Tape, and above-mentioned forecast for market tendency value is with void
Line is shown in sentiment undertone.
The above-mentioned forecast for market tendency value of the analysis target or comprehensive forecast for market tendency value, which can be pushed or be requested, is sent to use
The interface that family is specified.
Analysis target is subjected to Plate division and predicts the analysis target in each plate market in the block, and by described each
A plate forecast for market tendency value carries out Similarity-Weighted and is averaged, and takes Similarity-Weighted average value pre- as the market of the analysis target
Measured value makes moderate progress to the Stability and veracity of forecast for market tendency long-term in analysis target, and it is more accurate, quick to carry out
Prediction.
Embodiment two
The present embodiment is considered as being further improved to the Value investment analysis method based on machine learning described in embodiment one
Flow with reference to figure 2, shows the flow chart of Value investment analysis method of the embodiment of the present invention two based on machine learning, step
201 is identical to step 104 with the step 101 of embodiment one to step 204, and details are not described herein.It is different from embodiment one it
It is in the present embodiment, which further includes, establishes interaction relation step 205：
It obtains with reference to target, by analysis target with establishing interaction relation with reference to target；
It calculates and compares the reference target with analyzing the financial similarity of target, if the financial similarity difference of the two is more than
Threshold value then provides the warning prompt information of analysis target, and investor is prompted to analyze target and is existed with the financial factor with reference to target
Difference, difference therein include difference, ratio or relative value.
The above-mentioned screening technique with reference to target includes the following steps：
Selection is with reference to the range of target：Including all targets in the plate with analyzing target similarity maximum as the first mark
's；
It obtains with reference to target, calculate the first target and analyzes the plate similarity of target, it is more than threshold value or phase to take similarity
It is used as like the first maximum target of degree with reference to target, wherein, the plate similarity is by extracting first target and analysis
The plate keyword of target and the similarity acquisition for calculating the keyword.
The financial similarity is by calculating the similarity with reference to target and the financial factor of the analysis target
Obtain, the similarity calculating method include standardization Euclidean distance, beat up it is several in distance, manhatton distance, pearson correlation
Property, however it is not limited to this.The above-mentioned finance factor both includes preceding quarter earnings per share, net profit on sales rate, net assets income ratio, per share
Management road net profit, operating income, main management industry year-on-year growth rate, prime operating revenue sequential growth rate, operating profit, net profit
One in profit, total profit, net profit year-on-year growth rate, net profit sequential growth rate, cash flow, asset-liability ratio or
Multinomial combination or it is multinomial between arithmetic generate the factor.Preferably, with reference to target and season in the analysis target
Net profit, operating income, business growth rate, profit growth rate, asset-liability ratio are spent as the financial factor and calculate the phase of the two
Financial similarity is obtained like degree.
By the way that the interaction relation of financial data will be established with the analysis target with reference to target, by comparing the finance of the two
Correlation or similarity come the financial index of monitoring analysis target, the analysis mark that investor's early warning can be supplied to be investigated in this way
The possibility or it is possible that the suspicion played tricks that have large change on financial index.
Step 206, result is shown.
Show analysis result, specifically, include display analysis each plate of target forecast for market tendency value, weighted value and
Comprehensive forecast for market tendency value, above-mentioned display mode include single On the Tape or superposition On the Tape, and above-mentioned forecast for market tendency value is with void
Line is shown in sentiment undertone.
The above-mentioned forecast for market tendency value of the analysis target or comprehensive forecast for market tendency value, which can be pushed or be requested, is sent to use
The interface that family is specified.
Or above-mentioned Financial Crisis Prediction information is pushed to user.
Analysis target is subjected to Plate division and predicts the analysis target in each plate market in the block, and by described each
A plate forecast for market tendency value carries out Similarity-Weighted and is averaged, and takes Similarity-Weighted average value pre- as the market of the analysis target
Measured value makes moderate progress to the Stability and veracity of forecast for market tendency long-term in analysis target, and it is more accurate, quick to carry out
Prediction.
Embodiment three
The present invention also provides a kind of computing device, including：
One or more processors；
Memory；And
One or more programs, wherein one or more of programs are stored in the memory and are configured as by institute
One or more processors execution is stated, one or more of programs include performing following method：
(1) analysis target is subjected to Plate division, calculates analysis target and the similarity of the plate；
(2) structure builds neural network model based on neural network theory, and the neural network module includes at least input
Layer, hidden layer and output layer, the input layer are the linear input layer of multidimensional；
(3) using the financial factor in a period and plate similarity in the analysis target, as input layer parameter, prediction should
Analyze market of the target lower period in each plate；
(4) it carries out Similarity-Weighted to each plate forecast for market tendency value to be averaged, takes Similarity-Weighted average value conduct
The forecast for market tendency value of the analysis target；
(5) result is shown.
Example IV
The present invention also provides a kind of computer storage media, the storage medium is stored with one or more programs, described
One or more programs include instruction, and described instruction is when executed by a computing apparatus so that the computing device performs such as lower section
Method：
(1) analysis target is subjected to Plate division, calculates analysis target and the similarity of the plate；
(2) structure builds neural network model based on neural network theory, and the neural network module includes at least input
Layer, hidden layer and output layer, the input layer are the linear input layer of multidimensional；
(3) using the financial factor in a period and plate similarity in the analysis target, as input layer parameter, prediction should
Analyze market of the target lower period in each plate；
(4) it carries out Similarity-Weighted to each plate forecast for market tendency value to be averaged, takes Similarity-Weighted average value conduct
The forecast for market tendency value of the analysis target；
(5) result is shown.
Method, apparatus or module described in above-described embodiment can specifically be realized or by computer chip or entity by having
The product of certain function realizes, wherein, a kind of typical equipment is computer.Specifically, computer can be individual calculus
Machine, laptop computer, cellular phone, camera phone, smart phone, personal digital assistant, media player, is led at server
Any equipment in equipment, electronic mail equipment, game console platform, tablet computer, wearable device or these equipment of navigating
Combination.
It will be understood by those skilled in the art that the embodiment of the present invention can providing method, system or computer program product.
Therefore, the embodiment in terms of complete hardware embodiment, complete software embodiment or combination software and hardware can be used in the present invention
Form.It can be used and deposit moreover, one or more computers for wherein including computer usable program code can be used in the present invention
The shape of computer program product that storage media is implemented on (including but not limited to magnetic disk storage, CD, ROM, optical memory etc.)
Formula.
The foregoing is merely the embodiment of the present invention, are not intended to restrict the invention.To those skilled in the art,
The invention may be variously modified and varied.All any modifications made within spirit and principles of the present invention, equivalent replacement,
Improve etc., it should be included within scope of the presently claimed invention.
Claims (10)
1. a kind of Value investment analysis method based on machine learning, suitable for being performed in computing device, which is characterized in that including
Following steps：
(1) analysis target is subjected to Plate division, calculates analysis target and the similarity of the plate；
(2) structure builds neural network model based on neural network theory, and the neural network module includes at least input layer, hidden
Layer and output layer are hidden, the input layer is the linear input layer of multidimensional；
(3) analysis as input layer parameter, is predicted using the financial factor in a period and plate similarity in the analysis target
A target lower period is in the market of each plate；
(4) it carries out Similarity-Weighted to each plate forecast for market tendency value to be averaged, take described in the conduct of Similarity-Weighted average value
Analyze the forecast for market tendency value of target；
(5) result is shown.
2. the Value investment analysis method according to claim 1 based on machine learning, which is characterized in that the step
(1) in, an analysis target can be divided into multiple plates, and the analysis target division method of slab includes：
The analysis basic side information of target and technological side data information are obtained, is extracted in basic side information and technological side data information
Keyword；
Merge the word that synonym or similarity in the keyword meet threshold value, generate plate characteristic word, and crucial from plate
It is found in dictionary with word similar in plate characteristic word as plate title, the plate title is belonging to the analysis target
Plate；
The computational methods of the analysis target and the similarity of affiliated plate include：
Extraction and analysis target basic side information and technological side data information keyword；
Calculate the similarity of the keyword and affiliated plate characteristic word.
3. the Value investment analysis method according to claim 1 based on machine learning, which is characterized in that the step
(3) in, the financial factor includes earnings per share, net profit on sales rate, net assets income ratio, per share management road net profit
Profit, operating income, main management industry year-on-year growth rate, prime operating revenue sequential growth rate, operating profit, net profit, total profit, net profit
Moisten year-on-year growth rate, net profit sequential growth rate, cash flow, the one or more combination in asset-liability ratio or more
The factor that arithmetic between generates；
The market include opening price, closing price, highest price, lowest price, exchange hand, transaction value, turnover rate, circulation value, total
One or more combinations in market value, shareholder's number or it is multinomial between the market factor that generates of arithmetic.
4. the Value investment analysis method according to claim 1 based on machine learning, which is characterized in that the step
(3) it in, before the financial factor parameter of input, further includes：The step of financial factor is clustered, cluster generation debt paying ability because
Son, the operation ability factor, the profitability factor, enterprise development capability facfor.
5. the Value investment analysis method according to claim 1 based on machine learning, which is characterized in that the step
(4) in, the forecast for market tendency value of the analysis target：
Wherein, N is the plate quantity belonging to the analysis target,J-th of predicted value of the plate market for the i-th plate, xiFor
The analysis target and the similarity of affiliated plate.
6. the Value investment analysis method according to claim 1 based on machine learning, which is characterized in that in the step
(4) it after, before step (5), further includes：
It obtains with reference to target, the analysis target is established into interaction relation with described with reference to target；
The similarity difference with reference to described in target with the financial similarity of the analysis target and comparison is calculated, if the wealth of the two
Similarity of being engaged in difference is more than threshold value, then provides the warning prompt information of analysis target, wherein, the difference includes difference, ratio
Value or relative value.
7. the Value investment analysis method according to claim 6 based on machine learning, which is characterized in that described with reference to mark
Screening technique include：
Selection is with reference to the range of target：Including all targets in the plate with analyzing target similarity maximum as the first target；
It obtains with reference to target：It calculates the first target and analyzes the plate similarity of target, it is more than threshold value or similarity to take similarity
The first maximum target is used as with reference to target, wherein, the plate similarity is by extracting first target and analysis target
Plate keyword and calculate the keyword similarity obtain.
8. the Value investment analysis method according to claim 6 based on machine learning, which is characterized in that the finance
Similarity is obtained by calculating the similarity with reference to target and the financial factor of the analysis target, the similarity calculation
Method include standardization Euclidean distance, beat up it is several in distance, manhatton distance, Pearson correlation.
9. a kind of computing device, including：
One or more processors；
Memory；And
One or more programs, wherein one or more of programs are stored in the memory and are configured as by described one
A or multiple processors perform, and one or more of programs include performing according in claim 1-8 the methods
The instruction of either method.
10. a kind of computer storage media, the storage medium is stored with one or more programs, one or more of programs
Including instruction, described instruction is when executed by a computing apparatus so that the computing device performs the side according to claim 1-8
Either method in method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711402200.XA CN108197729A (en) | 2017-12-22 | 2017-12-22 | Value investment analysis method, equipment and storage medium based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711402200.XA CN108197729A (en) | 2017-12-22 | 2017-12-22 | Value investment analysis method, equipment and storage medium based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108197729A true CN108197729A (en) | 2018-06-22 |
Family
ID=62583651
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711402200.XA Pending CN108197729A (en) | 2017-12-22 | 2017-12-22 | Value investment analysis method, equipment and storage medium based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108197729A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110930256A (en) * | 2019-09-30 | 2020-03-27 | 北京九章云极科技有限公司 | Quantitative analysis method and quantitative analysis system |
| CN112967122A (en) * | 2021-02-05 | 2021-06-15 | 马晓华 | Financial management system |
| CN114549195A (en) * | 2021-12-31 | 2022-05-27 | 上海华鑫股份有限公司 | Market value reevaluation analysis method based on industry chain map and visual analysis system |
-
2017
- 2017-12-22 CN CN201711402200.XA patent/CN108197729A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110930256A (en) * | 2019-09-30 | 2020-03-27 | 北京九章云极科技有限公司 | Quantitative analysis method and quantitative analysis system |
| CN112967122A (en) * | 2021-02-05 | 2021-06-15 | 马晓华 | Financial management system |
| CN114549195A (en) * | 2021-12-31 | 2022-05-27 | 上海华鑫股份有限公司 | Market value reevaluation analysis method based on industry chain map and visual analysis system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Huang et al. | Using social media mining technology to improve stock price forecast accuracy | |
| Liu et al. | Combining enterprise knowledge graph and news sentiment analysis for stock price prediction | |
| US10614073B2 (en) | System and method for using data incident based modeling and prediction | |
| Chen et al. | An alternative model for the analysis of detecting electronic industries earnings management using stepwise regression, random forest, and decision tree | |
| Liu et al. | Technology and cryptocurrency valuation: Evidence from machine learning | |
| Lu et al. | Credit rating change modeling using news and financial ratios | |
| Stanisic et al. | Corporate bankruptcy prediction in the Republic of Serbia | |
| CN108197729A (en) | Value investment analysis method, equipment and storage medium based on machine learning | |
| Ruyu et al. | A comparison of credit rating classification models based on spark-evidence from lending-club | |
| Liu et al. | Application of hierarchical clustering in tax inspection case-selecting | |
| Zhang et al. | Author impact: Evaluations, predictions, and challenges | |
| Keller et al. | Trading on cryptocurrency markets: Analyzing the behavior of bitcoin investors | |
| Curme et al. | Quantifying the diversity of news around stock market moves | |
| CN112102006A (en) | Target customer acquisition method, target customer search method and target customer search device based on big data analysis | |
| Colapinto et al. | Goal programming for financial portfolio management: a state-of-the-art review | |
| Navarro et al. | Stock market optimization amidst the COVID-19 pandemic: Technical analysis, K-means algorithm, and mean-variance model (TAKMV) approach | |
| Li et al. | Optimization of investment strategies through machine learning | |
| Park et al. | Hot topic detection in central bankers’ speeches | |
| Awad et al. | Using data mining tools to prediction of going concern on auditor opinion-empirical study in iraqi commercial | |
| Medianovskyi et al. | Interpretable machine learning for SME financial distress prediction | |
| Wang et al. | Application of Data Mining Technology in Financial Data Analysis Methods Under the Background of Big Data | |
| Zhang et al. | Financial fraud identification based on stacking ensemble learning algorithm: Introducing MD&A text information | |
| Baxter et al. | Predictive model of insolvency risk for Australian corporations | |
| Sayadi et al. | Prediction-based portfolio optimization model for Iran’s oil dependent stocks using data mining methods | |
| Cui | Research on Credit Risk Control of Commercial Banks Based on Data Mining Technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180622 |