CN107230151A - Stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium - Google Patents
Stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium Download PDFInfo
- Publication number
- CN107230151A CN107230151A CN201710391370.6A CN201710391370A CN107230151A CN 107230151 A CN107230151 A CN 107230151A CN 201710391370 A CN201710391370 A CN 201710391370A CN 107230151 A CN107230151 A CN 107230151A
- Authority
- CN
- China
- Prior art keywords
- data
- training
- stock
- input data
- data set
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/06—Asset management; Financial planning or analysis
Abstract
The invention discloses a kind of stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium.Wherein method includes：The corresponding stock certificate data value of multiple stock certificate datas is extracted as training input data, stock rising trend data value or stock depreciation tendency data value are extracted as training output data, the training input data set of the training input data composition of a couple of days and the training output data set of training output data composition is obtained；Training input data set is inputted into multilayer convolutional network, obtains predicting output data set, training obtains the network model parameter of multilayer convolutional network；Stock test data is obtained, the corresponding stock certificate data value of multiple stock certificate datas is extracted as test input data, test input data is inputted into multilayer convolutional network, corresponding prediction output data is obtained.The present invention is using big data fitting, it is established that huge and without the deep learning model of strong event driven stock certificate data, can realize effective prediction to stock future ups and downs situation.
Description
Technical field
The present invention relates to technical field of data processing, and in particular to a kind of stock certificate data Forecasting Methodology and device, electronics are set
Standby, computer-readable storage medium.
Background technology
Stock is as a kind of way to manage money of high repayment, and it is favored by many investors, but at the same time, stock
High risk also counteracts that a part of potential investor.Due to influenceing the factor of stock price numerous and specialty of finance data
Stronger, the common investor of property, the investor especially newly got started is unfamiliar with market, to data deficiency sensitiveness, therefore can not
The ups and downs trend of rational prediction stock.
During the embodiment of the present invention is realized, inventor, which has found to realize in the prior art to utilize, influences stock price
Multinomial historical data forecast model is set up to predict the ups and downs of stock, more clearly guided to be provided for investor.
The content of the invention
In view of the above problems, it is proposed that the present invention so as to provide one kind overcome above mentioned problem or at least in part solve on
State the stock certificate data Forecasting Methodology and corresponding stock certificate data prediction meanss, electronic equipment, computer-readable storage medium of problem.
According to an aspect of the invention, there is provided a kind of stock certificate data Forecasting Methodology, this method includes：
The stock historical data in a couple of days is gathered in units of day, extracts and presets in daily stock historical data
The corresponding stock certificate data value of multiple stock certificate datas as daily training input data, in stock historical data every other day
Stock rising trend data value or stock depreciation tendency data value are extracted as the daily corresponding training output data of training input data, most
The training of the training input data set and the training output data of a couple of days composition of the training input data composition of a couple of days is obtained eventually
Output data set；
Training input data set is inputted into multilayer convolutional network, prediction corresponding with training input data set is obtained defeated
Go out data set, multilayer is obtained using prediction output data set corresponding with training input data set and training output data set training
The network model parameter of convolutional network；
Multiple number of share of stocks set in advance are extracted in stock test data on the day of acquisition, the stock test data on the day of
According to test input data of the corresponding stock certificate data value of item as the same day, the test input data on the same day is inputted to training and obtained
Multilayer convolutional network in, obtain the corresponding prediction output data every other day with the test input data on the same day, prediction every other day
Output data is specially stock rising trend data value or stock depreciation tendency data value.
According to another aspect of the present invention there is provided a kind of stock certificate data prediction meanss, the device includes：
Acquisition module, suitable for gathering the stock historical data in a couple of days in units of day, in daily stock historical data
The middle corresponding stock certificate data value of multiple stock certificate datas set in advance of extracting is as daily training input data, every other day
Stock rising trend data value is extracted in stock historical data or stock depreciation tendency data value is corresponding as daily training input data
Output data is trained, training input data set and the training of a couple of days output of the training input data composition of a couple of days is finally given
The training output data set of data composition；
Training module, suitable for training input data set is inputted into multilayer convolutional network, is obtained and training input data
Collect corresponding prediction output data set, utilize prediction output data set corresponding with training input data set and training output data
Training gets the network model parameter of multilayer convolutional network；
Prediction module, suitable for acquisition on the day of stock test data, extract in the stock test data on the day of and set in advance
The corresponding stock certificate data value of fixed multiple stock certificate datas as the same day test input data, by the test input data on the same day
Input in the multilayer convolutional network obtained to training, obtain prediction output number every other day corresponding with the test input data on the same day
According to prediction output data every other day is specially stock rising trend data value or stock depreciation tendency data value.
According to another aspect of the present invention there is provided a kind of electronic equipment, including：Processor, memory, communication interface and
Communication bus, processor, memory and communication interface complete mutual communication by communication bus；Memory is used to deposit extremely
A few executable instruction, executable instruction makes the corresponding operation of computing device stock certificate data Forecasting Methodology.
According to another aspect of the present invention there is provided a kind of computer-readable storage medium, at least one is stored with storage medium
Executable instruction, executable instruction makes the corresponding operation of computing device stock certificate data Forecasting Methodology.
It is every by collection according to the stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium of the present invention
It stock certificate data gathers daily advance versus decline value as training output data as daily training input data, when
Between on, daily training input data is corresponding with the training output data of second day；Input training input data set to multilayer is rolled up
In product network, multilayer convolution is obtained using prediction output data set corresponding with training input data set and training output data set
The network model parameter of network；Obtain stock test data to be input in multilayer convolutional network, prediction is obtained and test input number
According to corresponding prediction output data., can be big to data volume and no strong according to the stock certificate data Forecasting Methodology of the present embodiment
Event driven stock certificate data fits deep learning model using big data, and many factors for influenceing stock price ups and downs are comprehensive
Close and consider, be achieved in the beneficial effect of Accurate Prediction advance versus decline situation.
Described above is only the general introduction of technical solution of the present invention, in order to better understand the technological means of the present invention,
And can be practiced according to the content of specification, and in order to allow above and other objects of the present invention, feature and advantage can
Become apparent, below especially exemplified by the embodiment of the present invention.
Brief description of the drawings
By reading the detailed description of hereafter preferred embodiment, various other advantages and benefit is common for this area
Technical staff will be clear understanding.Accompanying drawing is only used for showing the purpose of preferred embodiment, and is not considered as to the present invention
Limitation.And in whole accompanying drawing, identical part is denoted by the same reference numerals.In the accompanying drawings：
Fig. 1 shows the flow chart of stock certificate data Forecasting Methodology according to an embodiment of the invention；
Fig. 2 shows the flow chart of stock certificate data Forecasting Methodology in accordance with another embodiment of the present invention；
Fig. 3 shows the schematic diagram of input data according to an embodiment of the invention and output data corresponding relation；
Fig. 4 shows the network model figure of two layers of convolutional network according to an embodiment of the invention；
Fig. 5 shows the functional block diagram of stock certificate data prediction meanss according to an embodiment of the invention；
Fig. 6 shows the functional block diagram of stock certificate data prediction meanss in accordance with another embodiment of the present invention；
Fig. 7 shows the structural representation of a kind of electronic equipment according to embodiments of the present invention.
Embodiment
The exemplary embodiment of the disclosure is more fully described below with reference to accompanying drawings.Although showing the disclosure in accompanying drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
Limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
Complete conveys to those skilled in the art.
Fig. 1 shows the flow chart of stock certificate data Forecasting Methodology according to an embodiment of the invention.As shown in figure 1, should
Method comprises the following steps：
Step S101, the stock historical data in a couple of days is gathered in units of day, is carried in daily stock historical data
The corresponding stock certificate data value of multiple stock certificate datas set in advance is taken as daily training input data, in stock every other day
Stock rising trend data value is extracted in historical data or stock depreciation tendency data value is used as the daily corresponding training of training input data
Output data, finally gives the training input data set and the training output data of a couple of days of the training input data composition of a couple of days
The training output data set of composition.
Specifically, from the historical data of certain stock, the data value that multiple data item are extracted using in units of day is used as one
The initial data of it training input data, wherein, data item is the factor of influence stock price ups and downs；And extract and training
The advance versus decline data value of corresponding second day of input data (every other day) is as training output data, if for example, the stock of second day
Ticket rise in price, then corresponding stock rising trend data value is 1, if the stock prices decline of second day, corresponding stock depreciation tendency
Data value is 0.The training input data of many days is extracted with same method and output data is trained, it is defeated by the above-mentioned training of many days
Enter data and training output data separately constitutes training input data set and training output data set.
Step S102, training input data set is inputted into multilayer convolutional network, is obtained and training input data set pair
The prediction output data set answered, utilizes prediction output data set corresponding with training input data set and training output data training
Get the network model parameter of multilayer convolutional network.
Specifically, input number of training input data set said extracted obtained as the first layer of multilayer convolutional network
According to the prediction output data set corresponding with training input data set and training output number obtained using multilayer convolutional network model
According to the error of collection, the parameter of adaptive adjustment multilayer convolutional network model, until error reaches default value, network mould now
Shape parameter has determined that the multilayer convolutional network model for being used to predict of the present embodiment.
Extract set in advance in step S103, the stock test data on the day of acquisition, the stock test data on the day of
The corresponding stock certificate data value of multiple stock certificate datas inputs the test input data on the same day as the test input data on the same day
In the multilayer convolutional network obtained to training, the corresponding prediction output data every other day with the test input data on the same day is obtained,
Prediction output data every other day is specially stock rising trend data value or stock depreciation tendency data value.
When needing prediction certain the stock ups and downs situation of second day, identical multiple number of share of stocks when collection is with for training
According to the corresponding stock certificate data value of item as test input data, and it is input in fixed multilayer convolutional network model, obtains
The prediction output data of second day.
Stock certificate data Forecasting Methodology provided in an embodiment of the present invention, by gathering, daily stock certificate data is used as daily instruction
Practice input data, gather daily advance versus decline value as training output data, in time, daily training input data with
The training output data correspondence of second day；Input training input data set inputs number into multilayer convolutional network using with training
According to the network model parameter for collecting corresponding prediction output data set with training output data set to obtain multilayer convolutional network；Obtain stock
Ticket test data is input in multilayer convolutional network, and prediction obtains prediction output data corresponding with test input data.According to
The stock certificate data Forecasting Methodology of the present embodiment, can be counted greatly and without strong event driven stock certificate data to data volume using big
According to deep learning model is fitted, many factors for influenceing stock price ups and downs are considered, Accurate Prediction stock is achieved in
The beneficial effect of ticket ups and downs situation.
Fig. 2 shows the flow chart of stock certificate data Forecasting Methodology in accordance with another embodiment of the present invention.As shown in Fig. 2
This method comprises the following steps：
Step S201, gathers stock historical data, and extract the training input data composition training input data set of a couple of days
Training output data composition training output data set every other day, wherein, training input data includes multinomial influence stock price
The corresponding value of the data item of ups and downs.
Specifically, multiple stock certificate datas include any multinomial in data below：Opening price, closing price, trading volume,
Highest price, 5 average daily lines, 5 daily turnovers, 10 average daily lines, 10 daily turnovers, turnover rate, Exponential Moving Average, simple shifting
Dynamic average line and weighted moving average line.
Because the cycle that the collection of some finance datas needs is longer, and limited, opened the set by exchange hour within one month
Number of days there was only 20 days or so, therefore when gathered data, the problems such as being considered as the time cycle.To gather exponential smoothing shifting
The data instance of dynamic average line, it needs the average value of the data of nearest 26 days.
Fig. 3 shows the schematic diagram of input data according to an embodiment of the invention and output data corresponding relation.Such as
Shown in Fig. 3, input data is the k such as the opening price of m days, closing price, the highest price extracted from the stock historical data collected
The data value of item data, forms m group input datas, and m group input datas may make up m*k input data matrix, Fig. 3
In, MACD is Exponential Moving Average, and WMA is weighted moving average line, the input data matrix that the input data is constituted
The test input data for assessing network model forecasting accuracy can also be used as training input data set during training；
Output data is the stock rising trend data value or depreciation tendency data value every other day of each group of data of correspondence.With the 1st group of data instance, carry
The data item k taken is that the input data length of group data of 64, i.e., the 1st is each data item on the 64, the 1st group of data corresponding same day
Data value be respectively x11, x12 ... x1n ... x164, the ups and downs case values of the stock every other day extracted use 0 table to fall
Show depreciation tendency data value.
Step S202, sets the convolution number of plies, pond specifications parameter and the initial network model parameter of multilayer convolutional network.
In the present embodiment, the convolution number of plies of multilayer convolutional network is two layers, and pond specifications parameter is 2*2.
Step S203, training input data set is converted to and inputted after the training input vector of default dimension to multilayer convolution
In network.
In the present embodiment, the vector for presetting dimension is four dimensional vectors.
So that the training input data of collection is the corresponding stock certificate data value of 64 stock certificate datas as an example, then the training of one day
Input data can regard 1*64 vector as, it is assumed that one group of training input data for [0.06999999999999984,
83.888888888888928,87.287342287342298,57.105025261063012,-
26.315789473684124,…..,127.86885245901655,58.223378529133605,
100427.45000000001,2.3728813559321882,9.0526006774772796], utilize python reshape
Function, four-dimensional vector x 1, specially x1=tf.reshape (x, [- 1,8,8,1]), wherein tf are converted into by training input data
For a function library of python language.
Corresponding one group of training input data every day that each group of training input data is concentrated all is converted to the four-dimension
Training input vector is input to the first layer of multilayer convolutional network.
Step S204, calculates the mistake of the corresponding prediction output data set of training input data set and training output data set
Difference, optimizes to meet preset error value using the error to the parameter of network model, obtains multilayer convolutional network.
Choose prediction output data set corresponding with training input data set and train the cross entropy of output data set to be instruction
Experienced loss function；Using gradient descent method, the network model parameter of multilayer convolutional network is optimized processing to meet damage
Lose function and reach preparatory condition.
Fig. 4 shows the network model figure of two layers of convolutional network according to an embodiment of the invention.As shown in figure 4,
The corresponding training input data of input data matrix is converted into after the training input vector of default dimension, will train input vector
Convolution is sought with the network model parameter w1 of the first layer of multilayer convolutional network, specific algorithm is：Y1=tf.nn.conv2d (x1,
W1)+b1, wherein x1 are the training input vector of default dimension, and w1 and b1 are network model parameter, and y1 is convolution results；Will
Convolution results carry out pond, i.e., y1 is taken to the pond in average or maximum, such as the present embodiment according to default pond specification
Specification is 2*2, exactly asks for average or maximum successively with 2*2 specification to y1, has carried out converging operation equivalent to y1, obtained
Y2 is output as to first layer convolution.
The output y2 of the first layer convolution models being input in the second layer of multilayer convolutional network with second layer convolution are joined
Number w2 asks progress pond after convolution, i.e., convolution results y3 is taken into average or maximum according to default pond specification, obtain second
Layer convolution is output as y4.
The output y4 of second layer convolution is output to a full articulamentum, i.e., y4 and a model parameter are carried out to w3 coupling
Close, specific algorithm is：Y5=y4*w3+b, wherein w3 and b are network model parameter, and y5 is coupling result, coupling result y5 warps
It is the input data matrix of the prediction output data, then m*k every other day of correspondence training input vector after the processing of softmax functions
Also there is corresponding m*1 output data matrix.The prediction output data every other day of the correspondence training input vector is exported with training
Data error value must reach preset error value, and the process of training could terminate, at the end of network model parameter determined that use
In the multilayer convolutional network model of prediction.
In the present embodiment, the loss function of the calculation error size of selection exports for the prediction of correspondence training input vector
The cross entropy of data and training output data, i.e. ,-∑ y-* ㏒ (y), wherein, y- trains the prediction output of input vector for correspondence
Data, y is training output data；When the result of the cross entropy does not meet preparatory condition, multilayer is optimized using gradient descent method
The network model parameter of convolutional network, until reach preparatory condition, that is, when finding out the network model parameter for making cross entropy minimum, instruction
White silk terminates, and obtains the network model parameter of multilayer convolutional network.
Step S205, by test input data be converted to default dimension test input vector input to train obtain it is many
In layer convolutional network, prediction obtains prediction output data corresponding with test input data.
When needing prediction certain the stock ups and downs situation of second day, identical multiple number of share of stocks when collection is with for training
According to the corresponding stock certificate data value of item as test input data, input data will be tested and be converted to the test input vector of the four-dimension simultaneously
It is input in fixed multilayer convolutional network model, obtains the prediction output data of second day.
Before the present embodiment step S205, the multilayer convolutional network model that can also be obtained to step S204 is commented
Estimate, further the accuracy rate and validity of checking model.
Stock certificate data Forecasting Methodology provided in an embodiment of the present invention, default vector is converted into by the training input data of extraction
It is input in multilayer convolutional network；Optimize the network of multilayer convolutional network using gradient descent method and according to the value of loss function
Model parameter；Pair determine multilayer convolutional network model be estimated, input prediction input data is predicted.According to this implementation
The stock certificate data Forecasting Methodology of example, can be fitted greatly and without strong event driven stock certificate data to data volume using big data
Go out deep learning model, many factors for influenceing stock price ups and downs are considered, Accurate Prediction advance versus decline is achieved in
The beneficial effect of situation.
Fig. 5 shows the functional block diagram of stock certificate data prediction meanss according to an embodiment of the invention.As shown in figure 3,
The device includes acquisition module 31, training module 32 and prediction module 33.
Acquisition module 31, suitable for gathering the stock historical data in a couple of days in units of day, in daily stock history number
According to the middle corresponding stock certificate data value of multiple stock certificate datas set in advance of extracting as daily training input data, every other day
Stock historical data in extract stock rising trend data value or stock depreciation tendency data value and be used as daily training input data correspondence
Training output data, finally give a couple of days training input data composition training input data set and the training of a couple of days it is defeated
Go out the training output data set of data composition.
Training module 32, suitable for training input data set is inputted into multilayer convolutional network, is obtained and training input number
According to corresponding prediction output data set is collected, prediction output data set corresponding with training input data set and training output number are utilized
The network model parameter of multilayer convolutional network is got according to training.
Prediction module 33, suitable for acquisition on the day of stock test data, extract in the stock test data on the day of advance
The test on the same day is inputted number by the corresponding stock certificate data value of multiple stock certificate datas of setting as the test input data on the same day
In the multilayer convolutional network obtained according to input to training, prediction output every other day corresponding with the test input data on the same day is obtained
Data, prediction output data every other day is specially stock rising trend data value or stock depreciation tendency data value.
Stock certificate data prediction meanss provided in an embodiment of the present invention, by gathering, daily stock certificate data is used as daily instruction
Practice input data, gather daily advance versus decline value as training output data, in time, daily training input data with
The training output data correspondence of second day；Input training input data set inputs number into multilayer convolutional network using with training
According to the network model parameter for collecting corresponding prediction output data set with training output data set to obtain multilayer convolutional network；Obtain stock
Ticket test data is input in multilayer convolutional network, and prediction obtains prediction output data corresponding with test input data.According to
The stock certificate data prediction meanss of the present embodiment, can be counted greatly and without strong event driven stock certificate data to data volume using big
According to deep learning model is fitted, many factors for influenceing stock price ups and downs are considered, Accurate Prediction stock is achieved in
The beneficial effect of ticket ups and downs situation.
Fig. 6 shows the functional block diagram of stock certificate data prediction meanss in accordance with another embodiment of the present invention.Such as Fig. 6 institutes
Show, Fig. 6 is in addition to including the modules shown in Fig. 5, in addition to initialization module 41.
Initialization module 41, the convolution number of plies, pond specifications parameter and initial network suitable for setting multilayer convolutional network
Model parameter.
In the present embodiment, the convolution number of plies of multilayer convolutional network is two layers, and pond specifications parameter is 2*2.
Acquisition module 31, suitable for gathering the stock historical data in a couple of days in units of day, in daily stock history number
According to the middle corresponding stock certificate data value of multiple stock certificate datas set in advance of extracting as daily training input data, every other day
Stock historical data in extract stock rising trend data value or stock depreciation tendency data value and be used as daily training input data correspondence
Training output data, finally give a couple of days training input data composition training input data set and the training of a couple of days it is defeated
Go out the training output data set of data composition.
Wherein, multiple stock certificate datas include any multinomial in data below：Opening price, closing price, trading volume, most
At high price, 5 average daily lines, 5 daily turnovers, 10 average daily lines, 10 daily turnovers, turnover rate, Exponential Moving Average, simple movement
Average line and weighted moving average line.
Training module 32, suitable for being inputted at most after training input data set is converted into the training input vector of default dimension
In layer convolutional network, prediction output data set corresponding with training input data set is obtained, using with training input data set pair
The prediction output data set and training output data set training answered obtain the network model parameter of multilayer convolutional network.
In the present embodiment, the vector for presetting dimension is four dimensional vectors.
In training process, training module 32 be further adapted for choose with training input data set it is corresponding predict output data set and
It is the loss function of training to train the cross entropy of output data set；Using gradient descent method, to the network mould of multilayer convolutional network
Shape parameter optimizes processing and reaches preparatory condition to meet loss function.
Prediction module 33, suitable for the test input vector for testing input data and being converted to default dimension is inputted to training
To multilayer convolutional network in, prediction is obtained with testing the corresponding prediction output data of input data.
Stock certificate data prediction meanss provided in an embodiment of the present invention, default vector is converted into by the training input data of extraction
It is input in multilayer convolutional network；Optimize the network of multilayer convolutional network using gradient descent method and according to the value of loss function
Model parameter；Pair determine multilayer convolutional network model be estimated, input prediction input data is predicted.According to this implementation
The stock certificate data prediction meanss of example, can be fitted greatly and without strong event driven stock certificate data to data volume using big data
Go out deep learning model, many factors for influenceing stock price ups and downs are considered, Accurate Prediction advance versus decline is achieved in
The beneficial effect of situation.
The embodiment of the present invention additionally provides a kind of nonvolatile computer storage media, the computer-readable storage medium storage
There is an at least executable instruction, the computer executable instructions can perform the stock certificate data prediction in above-mentioned any means embodiment
Method.
Fig. 7 shows the structural representation of a kind of electronic equipment according to embodiments of the present invention, the specific embodiment of the invention
Implementing for electronic equipment is not limited.
As shown in fig. 7, the electronic equipment can include：Processor (processor) 602, communication interface
(Communications Interface) 604, memory (memory) 606 and communication bus 608.
Wherein：
Processor 602, communication interface 604 and memory 606 complete mutual communication by communication bus 608.
Communication interface 604, communicates for the network element with miscellaneous equipment such as client or other servers etc..
Processor 602, for configuration processor 610, can specifically be performed in above-mentioned stock certificate data Forecasting Methodology embodiment
Correlation step.
Specifically, program 610 can include program code, and the program code includes computer-managed instruction.
Processor 602 is probably central processor CPU, or specific integrated circuit ASIC (Application
Specific Integrated Circuit), or it is arranged to implement one or more integrated electricity of the embodiment of the present invention
Road.The one or more processors that electronic equipment includes, can be same type of processors, such as one or more CPU；Also may be used
To be different types of processor, such as one or more CPU and one or more ASIC.
Memory 606, for depositing program 610.Memory 606 may include high-speed RAM memory, it is also possible to also include
Nonvolatile memory (non-volatile memory), for example, at least one magnetic disk storage.
Program 610 specifically can be used for so that processor 602 performs following operation：
The stock historical data in a couple of days is gathered in units of day, extracts and presets in daily stock historical data
The corresponding stock certificate data value of multiple stock certificate datas as daily training input data, in stock historical data every other day
Stock rising trend data value or stock depreciation tendency data value are extracted as the daily corresponding training output data of training input data, most
The training of the training input data set and the training output data of a couple of days composition of the training input data composition of a couple of days is obtained eventually
Output data set；
The training input data set is inputted into multilayer convolutional network, obtains corresponding with the training input data set
Prediction output data set, utilize it is described with training input data set it is corresponding prediction output data set and the training output number
The network model parameter of multilayer convolutional network is got according to training；
Stock test data on the day of acquisition, extracts many personal shares set in advance in the stock test data on the day of described
The corresponding stock certificate data value of ticket data as the same day test input data, by the test input data on the same day input to
Train in obtained multilayer convolutional network, obtain prediction output number every other day corresponding with the test input data on the same day
According to the prediction output data every other day is specially stock rising trend data value or stock depreciation tendency data value.
The multiple stock certificate data includes any multinomial in data below：Opening price, closing price, trading volume, most
At high price, 5 average daily lines, 5 daily turnovers, 10 average daily lines, 10 daily turnovers, turnover rate, Exponential Moving Average, simple movement
Average line and weighted moving average line.
Program 610 specifically can be also used for so that processor 602 performs following operation：
The training input data set is converted to and inputted after the training input vector of default dimension to the multilayer convolution
In network；
The test input vector that the test input data on the same day is converted into default dimension, which is inputted to described, trains
To multilayer convolutional network in.
The vector of the default dimension is four dimensional vectors.
Program 610 specifically can be also used for so that processor 602 performs following operation：
The convolution number of plies, pond specifications parameter and the initial network model parameter of multilayer convolutional network are set.
The convolution number of plies of the multilayer convolutional network is two layers, and the pond specifications parameter is 2*2.
Program 610 specifically can be also used for so that processor 602 performs following operation：
Choose the friendship with the corresponding prediction output data set of the training input data set and the training output data set
Pitch loss function of the entropy for training；
Using gradient descent method, the network model parameter of multilayer convolutional network is optimized processing to meet loss function
Reach preparatory condition.
Implementing for each step may refer in above-mentioned application in stock certificate data Forecasting Methodology embodiment in program 610
Corresponding description, will not be described here in corresponding steps and unit.It is apparent to those skilled in the art that, to retouch
The convenience stated is with succinctly, and the equipment of foregoing description and the specific work process of module may be referred in preceding method embodiment
Corresponding process is described, and will not be repeated here.
Algorithm and display be not inherently related to any certain computer, virtual system or miscellaneous equipment provided herein.
Various general-purpose systems can also be used together with based on teaching in this.As described above, construct required by this kind of system
Structure be obvious.In addition, the present invention is not also directed to any certain programmed language.It is understood that, it is possible to use it is various
Programming language realizes the content of invention described herein, and the description done above to language-specific is to disclose this hair
Bright preferred forms.
In the specification that this place is provided, numerous specific details are set forth.It is to be appreciated, however, that the implementation of the present invention
Example can be put into practice in the case of these no details.In some instances, known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this description.
Similarly, it will be appreciated that in order to simplify the disclosure and help to understand one or more of each inventive aspect, exist
Above in the description of the exemplary embodiment of the present invention, each feature of the invention is grouped together into single implementation sometimes
In example, figure or descriptions thereof.However, the method for the disclosure should be construed to reflect following intention：It is i.e. required to protect
The application claims of shield features more more than the feature being expressly recited in each claim.More precisely, such as following
Claims reflect as, inventive aspect is all features less than single embodiment disclosed above.Therefore,
Thus the claims for following embodiment are expressly incorporated in the embodiment, wherein each claim is in itself
All as the separate embodiments of the present invention.
Those skilled in the art, which are appreciated that, to be carried out adaptively to the module in the equipment in embodiment
Change and they are arranged in one or more equipment different from the embodiment.Can be the module or list in embodiment
Member or component be combined into a module or unit or component, and can be divided into addition multiple submodule or subelement or
Sub-component.In addition at least some in such feature and/or process or unit exclude each other, it can use any
Combination is disclosed to all features disclosed in this specification (including adjoint claim, summary and accompanying drawing) and so to appoint
Where all processes or unit of method or equipment are combined.Unless expressly stated otherwise, this specification (including adjoint power
Profit is required, summary and accompanying drawing) disclosed in each feature can or similar purpose identical, equivalent by offer alternative features come generation
Replace.
Although in addition, it will be appreciated by those of skill in the art that some embodiments described herein include other embodiments
In included some features rather than further feature, but the combination of the feature of be the same as Example does not mean in of the invention
Within the scope of and form different embodiments.For example, in the following claims, times of embodiment claimed
One of meaning mode can be used in any combination.
The present invention all parts embodiment can be realized with hardware, or with one or more processor run
Software module realize, or realized with combinations thereof.It will be understood by those of skill in the art that can use in practice
Microprocessor or digital signal processor (DSP) realize one in stock certificate data prediction meanss according to embodiments of the present invention
The some or all functions of a little or whole parts.The present invention is also implemented as performing method as described herein
Some or all equipment or program of device (for example, computer program and computer program product).It is such to realize
The program of the present invention can be stored on a computer-readable medium, or can have the form of one or more signal.This
The signal of sample can be downloaded from internet website and obtained, and either provided or carried in any other form on carrier signal
For.
It should be noted that the present invention will be described rather than limits the invention for above-described embodiment, and ability
Field technique personnel can design alternative embodiment without departing from the scope of the appended claims.In the claims,
Any reference symbol between bracket should not be configured to limitations on claims.Word "comprising" is not excluded the presence of not
Element or step listed in the claims.Word "a" or "an" before element does not exclude the presence of multiple such
Element.The present invention can be by means of including the hardware of some different elements and coming real by means of properly programmed computer
It is existing.In if the unit claim of equipment for drying is listed, several in these devices can be by same hardware branch
To embody.The use of word first, second, and third does not indicate that any order.These words can be explained and run after fame
Claim.
The invention discloses：A1. a kind of stock certificate data Forecasting Methodology, it includes：
The stock historical data in a couple of days is gathered in units of day, extracts and presets in daily stock historical data
The corresponding stock certificate data value of multiple stock certificate datas as daily training input data, in stock historical data every other day
Stock rising trend data value or stock depreciation tendency data value are extracted as the daily corresponding training output data of training input data, most
The training of the training input data set and the training output data of a couple of days composition of the training input data composition of a couple of days is obtained eventually
Output data set；
The training input data set is inputted into multilayer convolutional network, obtains corresponding with the training input data set
Prediction output data set, utilize it is described with training input data set it is corresponding prediction output data set and the training output number
The network model parameter of multilayer convolutional network is got according to training；
Stock test data on the day of acquisition, extracts many personal shares set in advance in the stock test data on the day of described
The corresponding stock certificate data value of ticket data as the same day test input data, by the test input data on the same day input to
Train in obtained multilayer convolutional network, obtain prediction output number every other day corresponding with the test input data on the same day
According to the prediction output data every other day is specially stock rising trend data value or stock depreciation tendency data value.
A2. the method according to A1, wherein, the multiple stock certificate data includes any many in data below
：Opening price, closing price, trading volume, highest price, 5 average daily lines, 5 daily turnovers, 10 average daily lines, 10 daily turnovers, turnover rate,
Exponential Moving Average, simple Moving Average and weighted moving average line.
A3. the method according to A1 or A2, wherein, it is described to input the training input data set to multilayer convolution net
It is specially in network：The training input data set is converted to and inputted after the training input vector of default dimension to multilayer volume
In product network；
It is described to input the test input data on the same day in the multilayer convolutional network obtained to training specially：By institute
The test input vector that test input data on the day of stating is converted to default dimension is inputted to the multilayer convolution trained and obtained
In network.
A4. the method according to A3, wherein, the vector of the default dimension is four dimensional vectors.
A5. the method according to any one of A1-A4, wherein, it is described by the training input data set input to
Before in multilayer convolutional network, methods described also includes：Set the convolution number of plies of multilayer convolutional network, pond specifications parameter and
Initial network model parameter.
A6. the method according to A5, wherein, the convolution number of plies of the multilayer convolutional network is two layers, and the pondization is advised
Lattice parameter is 2*2.
A7. the method according to any one of A1-A6, described to utilize the prediction corresponding with training input data set
The network model parameter that output data set and training output data set training obtain multilayer convolutional network further comprises：
Choose the friendship with the corresponding prediction output data set of the training input data set and the training output data set
Pitch loss function of the entropy for training；
Using gradient descent method, the network model parameter of multilayer convolutional network is optimized processing to meet loss function
Reach preparatory condition.
The invention also discloses：B8. a kind of stock certificate data prediction meanss, it includes：
Acquisition module, suitable for gathering the stock historical data in a couple of days in units of day, in daily stock historical data
The middle corresponding stock certificate data value of multiple stock certificate datas set in advance of extracting is as daily training input data, every other day
Stock rising trend data value is extracted in stock historical data or stock depreciation tendency data value is corresponding as daily training input data
Output data is trained, training input data set and the training of a couple of days output of the training input data composition of a couple of days is finally given
The training output data set of data composition；
Training module, suitable for the training input data set is inputted into multilayer convolutional network, is obtained and the training
The corresponding prediction output data set of input data set, using it is described with training input data set it is corresponding prediction output data set and
The training output data set training obtains the network model parameter of multilayer convolutional network；
Prediction module, suitable for acquisition on the day of stock test data, extract pre- in the stock test data on the day of described
The corresponding stock certificate data value of multiple stock certificate datas for first setting as the same day test input data, by the test on the same day
Input data is inputted in the multilayer convolutional network obtained to training, obtains corresponding every other day with the test input data on the same day
Prediction output data, the prediction output data every other day is specially stock rising trend data value or stock depreciation tendency data value.
B9. the device according to B8, wherein, the multiple stock certificate data includes any many in data below
：Opening price, closing price, trading volume, highest price, 5 average daily lines, 5 daily turnovers, 10 average daily lines, 10 daily turnovers, turnover rate,
Exponential Moving Average, simple Moving Average and weighted moving average line.
B10. the device according to B8 or B9, wherein, the training module is further adapted for：The training is inputted into number
Inputted after being converted to the training input vector of default dimension according to collection into the multilayer convolutional network；
The prediction module is further adapted for：The test that the test input data on the same day is converted into default dimension is defeated
Incoming vector is inputted into the multilayer convolutional network trained and obtained.
B11. the device according to B10, wherein, the vector of the default dimension is four dimensional vectors.
B12. the device according to any one of B8-B11, wherein, described device also includes：Initialization module, is suitable to
The convolution number of plies, pond specifications parameter and the initial network model parameter of multilayer convolutional network are set.
B13. the device according to B12, wherein, the convolution number of plies of the multilayer convolutional network is two layers, the pond
Specifications parameter is 2*2.
B14. the device according to any one of B8-B13, the training module is further adapted for：Choose and the instruction
Practice loss function of the cross entropy of the corresponding prediction output data set of input data set and the training output data set for training；
Using gradient descent method, processing is optimized to the network model parameter of multilayer convolutional network with meet loss function reach it is default
Condition.
The invention also discloses：C15. a kind of electronic equipment, including：Processor, memory, communication interface and communication are total
Line, the processor, the memory and the communication interface complete mutual communication by the communication bus；
The memory is used to deposit an at least executable instruction, and the executable instruction makes the computing device such as
The corresponding operation of stock certificate data Forecasting Methodology any one of A1-A7.
The invention also discloses：D16. being stored with a kind of computer-readable storage medium, the storage medium at least one can hold
Row instruction, the executable instruction makes stock certificate data Forecasting Methodology pair of the computing device as any one of A1-A7
The operation answered.
Claims (10)
1. a kind of stock certificate data Forecasting Methodology, it includes：
The stock historical data in a couple of days is gathered in units of day, extracts set in advance many in daily stock historical data
The corresponding stock certificate data value of individual stock certificate data is extracted as daily training input data in stock historical data every other day
Stock rising trend data value or stock depreciation tendency data value are final to obtain as the daily corresponding training output data of training input data
The training output of the training input data set and the training output data of a couple of days composition that are constituted to the training input data of a couple of days
Data set；
The training input data set is inputted into multilayer convolutional network, obtains corresponding pre- with the training input data set
Output data set is surveyed, the prediction output data set corresponding with training input data set and the training output data set is utilized
Training obtains the network model parameter of multilayer convolutional network；
Stock test data on the day of acquisition, extracts multiple number of share of stocks set in advance in the stock test data on the day of described
According to test input data of the corresponding stock certificate data value of item as the same day, the test input data on the same day is inputted to training
In obtained multilayer convolutional network, the corresponding prediction output data every other day with the test input data on the same day, institute are obtained
The prediction output data stated every other day is specially stock rising trend data value or stock depreciation tendency data value.
2. according to the method described in claim 1, wherein, the multiple stock certificate data includes any many in data below
：Opening price, closing price, trading volume, highest price, 5 average daily lines, 5 daily turnovers, 10 average daily lines, 10 daily turnovers, turnover rate,
Exponential Moving Average, simple Moving Average and weighted moving average line.
3. method according to claim 1 or 2, wherein, it is described to input the training input data set to multilayer convolution
It is specially in network：The training input data set is converted to and inputted after the training input vector of default dimension to the multilayer
In convolutional network；
It is described to input the test input data on the same day in the multilayer convolutional network obtained to training specially：Described it will work as
The test input vector that it test input data is converted to default dimension is inputted to the multilayer convolutional network trained and obtained
In.
4. method according to claim 3, wherein, the vector of the default dimension is four dimensional vectors.
5. the method according to any one of claim 1-4, wherein, it is described by the training input data set input to
Before in multilayer convolutional network, methods described also includes：Set the convolution number of plies of multilayer convolutional network, pond specifications parameter and
Initial network model parameter.
6. method according to claim 5, wherein, the convolution number of plies of the multilayer convolutional network is two layers, the pond
Specifications parameter is 2*2.
7. the method according to any one of claim 1-6, described using described corresponding pre- with training input data set
Survey output data set and the training output data set trains the network model parameter for obtaining multilayer convolutional network to further comprise：
Choose the cross entropy with the corresponding prediction output data set of the training input data set and the training output data set
For the loss function of training；
Using gradient descent method, processing is optimized to the network model parameter of multilayer convolutional network and reached with meeting loss function
Preparatory condition.
8. a kind of stock certificate data prediction meanss, it includes：
Acquisition module, suitable for gathering the stock historical data in a couple of days in units of day, is carried in daily stock historical data
The corresponding stock certificate data value of multiple stock certificate datas set in advance is taken as daily training input data, in stock every other day
Stock rising trend data value is extracted in historical data or stock depreciation tendency data value is used as the daily corresponding training of training input data
Output data, finally gives the training input data set and the training output data of a couple of days of the training input data composition of a couple of days
The training output data set of composition；
Training module, suitable for the training input data set is inputted into multilayer convolutional network, obtains inputting with the training
The corresponding prediction output data set of data set, utilizes the prediction output data set and described corresponding with training input data set
Training output data set training obtains the network model parameter of multilayer convolutional network；
Prediction module, suitable for acquisition on the day of stock test data, extract and set in advance in the stock test data on the day of described
The fixed corresponding stock certificate data value of multiple stock certificate datas inputs the test on the same day as the test input data on the same day
Data input obtains every other day pre- corresponding with the test input data on the same day into the obtained multilayer convolutional network of training
Output data is surveyed, the prediction output data every other day is specially stock rising trend data value or stock depreciation tendency data value.
9. a kind of electronic equipment, including：Processor, memory, communication interface and communication bus, the processor, the storage
Device and the communication interface complete mutual communication by the communication bus；
The memory is used to deposit an at least executable instruction, and the executable instruction makes the computing device such as right will
Ask the corresponding operation of the stock certificate data Forecasting Methodology any one of 1-7.
10. be stored with an at least executable instruction, the executable instruction in a kind of computer-readable storage medium, the storage medium
Make the corresponding operation of stock certificate data Forecasting Methodology of the computing device as any one of claim 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710391370.6A CN107230151A (en) | 2017-05-27 | 2017-05-27 | Stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710391370.6A CN107230151A (en) | 2017-05-27 | 2017-05-27 | Stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107230151A true CN107230151A (en) | 2017-10-03 |
Family
ID=59933834
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710391370.6A Pending CN107230151A (en) | 2017-05-27 | 2017-05-27 | Stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107230151A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108876613A (en) * | 2018-06-06 | 2018-11-23 | 东莞市波动赢机器人科技有限公司 | Deep learning method, electronic equipment and the computer storage medium of transaction machine people |
| CN111192144A (en) * | 2020-01-03 | 2020-05-22 | 湖南工商大学 | Financial data prediction method, device, equipment and storage medium |
-
2017
- 2017-05-27 CN CN201710391370.6A patent/CN107230151A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108876613A (en) * | 2018-06-06 | 2018-11-23 | 东莞市波动赢机器人科技有限公司 | Deep learning method, electronic equipment and the computer storage medium of transaction machine people |
| CN111192144A (en) * | 2020-01-03 | 2020-05-22 | 湖南工商大学 | Financial data prediction method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cavalli et al. | CNN-based multivariate data analysis for bitcoin trend prediction | |
| Xie et al. | Variable selection methods for probabilistic load forecasting: Empirical evidence from seven states of the united states | |
| Sayavong et al. | Research on stock price prediction method based on convolutional neural network | |
| Rathnayaka et al. | A hybrid statistical approach for stock market forecasting based on artificial neural network and ARIMA time series models | |
| CN111695731B (en) | Load prediction method, system and equipment based on multi-source data and hybrid neural network | |
| CN104424598A (en) | Cash demand quantity predicating device and method | |
| CN107833214A (en) | Video definition detection method, device, computing device and computer-readable storage medium | |
| CN107895172A (en) | Utilize the method, apparatus and computing device of image information detection anomalous video file | |
| CN107230151A (en) | Stock certificate data Forecasting Methodology and device, electronic equipment, computer-readable storage medium | |
| Pedersen et al. | Optimizing a condition-based maintenance policy by taking the preferences of a risk-averse decision maker into account | |
| CN114943565A (en) | Electric power spot price prediction method and device based on intelligent algorithm | |
| Liang et al. | The analysis of global RMB exchange rate forecasting and risk early warning using ARIMA and CNN model | |
| CN107729873A (en) | Special object detection method, device and computing device in video file | |
| CN116707331B (en) | Inverter output voltage high-precision adjusting method and system based on model prediction | |
| CN112785057A (en) | Component prediction method, device, equipment and storage medium based on exponential smoothing | |
| Periketi et al. | LSTM and Prophet Model Fusion Framework for Bitcoin Candlestick Visualization and Price Prediction Forecasting | |
| Tomohiro et al. | Proposal of variable sequential sampling plan having desired operating characteristics indexed by quality loss | |
| CN108710994A (en) | Investment share-selecting method, device and storage medium based on the public sentiment factor | |
| CN113360416A (en) | Test data batch generation method and device | |
| Yeganegi et al. | Modeling e-Customer Relationship Management, Case Study: Banking Systems | |
| CN111768290A (en) | Method and device for determining risk weight coefficient of service | |
| Chauhan et al. | Study of Linear Regression Prediction Model for American Stock Market Prediction | |
| CN108269004A (en) | Life of product analysis method and terminal device | |
| Ulumuddin et al. | Bitcoin Price Prediction Using Long Short Term Memory (LSTM): Bitcoin Price Prediction Using Long Short Term Memory (LSTM) | |
| Putra et al. | Business Process Analysis and Modeling Using the Business Process Improvement Framework at the Internal Quality Assurance STMIK STIKOM Indonesia |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171003 |