CN109636017A - A kind of financial transaction price expectation method, apparatus, medium and equipment - Google Patents
A kind of financial transaction price expectation method, apparatus, medium and equipment Download PDFInfo
- Publication number
- CN109636017A CN109636017A CN201811442001.6A CN201811442001A CN109636017A CN 109636017 A CN109636017 A CN 109636017A CN 201811442001 A CN201811442001 A CN 201811442001A CN 109636017 A CN109636017 A CN 109636017A
- Authority
- CN
- China
- Prior art keywords
- data
- training
- financial transaction
- transaction
- lstm
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Marketing (AREA)
- Accounting & Taxation (AREA)
- Entrepreneurship & Innovation (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Technology Law (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The present invention provides a kind of financial transaction price expectation methods, comprising: acquires the historical data of financial transaction；Sample data is obtained from historical data, then the sample data is pre-processed, and obtains training data；Using the training data, it is based on LSTM-RNN model, training obtains transaction value prediction model；The data of financial transaction of the following any time period and/or any moment is predicted using the transaction value prediction model according to current transaction data.By the prediction model based on LSTM-RNN model foundation, compared to the prior art, the data of financial transaction of the following any time period and/or any moment can be more accurately predicted, improve user experience.
Description
Technical field
The present invention relates to technical field of data prediction, and in particular to a kind of financial transaction price expectation method, apparatus, medium
And equipment.
Background technique
With the development of internet industry, information technology is occupied an leading position, market direction of the securities market towards modernization
Development.Present Shanghai and Shenzhen listed company alreadys exceed thousands of families, however the income of financial investment to risk be often it is directly proportional, i.e.,
Investment return is higher, may risk risk is bigger.Therefore, financial transaction prediction research 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
Achieve biggish achievement, however, it is not difficult to find that these existing theory and methods be also have the defects that it is very big,
They be invariably using static method, qualitative description it is more, quantitative description lacks, and many factors that will affect financial transaction are cut
It splits and carrys out single analysis.Therefore, these limitations make these methods cannot in financial transaction price fluctuation changeable
Effectively and accurately hold the variation of financial transaction price.
Summary of the invention
For the defects in the prior art, the present invention provides a kind of financial transaction price expectation method, apparatus, medium and sets
It is standby, the data of financial transaction of the following any time period and/or any moment can be more accurately predicted, improve user experience.
In a first aspect, the present invention provides a kind of financial transaction price expectation methods, comprising:
Acquire the historical data of financial transaction；The sliding window of different predetermined time step-lengths is set, the sliding window is utilized
Mouth is slided in the historical data to obtain multiple window datas, is sampled to each window data, and each pre- timing is obtained
Between the corresponding sample data of step-length；
The sample data is pre-processed, training data is obtained；
Using the training data, it is based on LSTM-RNN model, training obtains transaction value prediction model；
According to current transaction data, using the transaction value prediction model, the following any time period and/or any is predicted
The data of financial transaction at moment.
Optionally, described that the sample data is pre-processed, obtain training data, comprising:
The sample data is cleaned；
Sample data after cleaning is normalized；
Classify to the data after normalized, obtains training data.
Optionally, the training data includes: training sample data, verify data and test data, described in the utilization
Training data, is based on LSTM-RNN model, and training obtains transaction value prediction model, comprising:
Utilize training sample data training LSTM-RNN model；
The LSTM-RNN model after training is modified using the verify data；
Revised LSTM-RNN model is tested using the test data；It is revised if being successfully tested
LSTM-RNN model is transaction value prediction model；If test crash reacquires verify data, to the LSTM- after training
RNN model re-starts amendment.
Optionally, further includes:
Using Adam algorithm, using the current transaction data, transaction value prediction model described in real-time optimization.
Optionally, the sample data of the financial transaction, comprising: financial product transaction value, exchange hour and transaction pair
As.
Optionally, further includes:
The data of financial transaction of the following any time period and/or any moment predicted is shown using visual means.
Second aspect, the present invention provide a kind of financial transaction price expectation device, comprising:
Data acquisition module, for acquiring the historical data of financial transaction；The sliding window of different predetermined time step-lengths is set
Mouthful, it is slided using the sliding window in the historical data to obtain multiple window datas, each window data is carried out
Sampling, obtains the corresponding sample data of each predetermined time step-length；
Data processing module obtains training data for pre-processing to the sample data；
Training module is based on LSTM-RNN model, training obtains transaction value and predicts mould for utilizing the training data
Type；
Prediction module is used for according to current transaction data, using the transaction value prediction model, when prediction is following any
Between section and/or the data of financial transaction of any moment.
The third aspect, the present invention provide a kind of computer readable storage medium, are stored thereon with computer program, the program
One of first aspect financial transaction price expectation method is realized when being executed by processor.
Fourth aspect, the present invention provide a kind of financial transaction price expectation equipment, comprising: memory, processor and storage
On a memory and the computer program that can run on a processor, the processor realize first aspect when executing described program
One of financial transaction price expectation method.
The present invention is by the prediction model based on LSTM-RNN model foundation, compared to the prior art, can be more accurate
The data of financial transaction of the following any time period and/or any moment is predicted on ground, improves user experience.
A kind of financial transaction price expectation device, a kind of computer readable storage medium and a kind of finance provided by the invention
The pre- measurement equipment of transaction value, it is having the same with a kind of above-mentioned financial transaction price expectation method for identical inventive concept
Beneficial effect.
Detailed description of the invention
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art
Embodiment or attached drawing needed to be used in the description of the prior art are briefly described.In all the appended drawings, similar element
Or part is generally identified by similar appended drawing reference.In attached drawing, each element or part might not be drawn according to actual ratio.
Fig. 1 is a kind of flow chart of financial transaction price expectation method provided in an embodiment of the present invention；
Fig. 2 is a kind of schematic diagram of financial transaction price expectation device provided in an embodiment of the present invention.
Specific embodiment
It is described in detail below in conjunction with embodiment of the attached drawing to technical solution of the present invention.Following embodiment is only used for
Clearly illustrate technical solution of the present invention, therefore be intended only as example, and cannot be used as a limitation and limit protection of the invention
Range.
It should be noted that unless otherwise indicated, technical term or scientific term used in this application should be this hair
The ordinary meaning that bright one of ordinary skill in the art are understood.
The present invention provides a kind of financial transaction price expectation method, apparatus, medium and equipment.With reference to the accompanying drawing to this
The embodiment of invention is illustrated.
Referring to FIG. 1, Fig. 1 is a kind of process for financial transaction price expectation method that the specific embodiment of the invention provides
Figure, a kind of financial transaction price expectation method provided in this embodiment, comprising:
Step S101: the historical data of financial transaction is acquired；The sliding window of different predetermined time step-lengths is set, institute is utilized
It states sliding window to slide in the historical data to obtain multiple window datas, each window data is sampled, is obtained
The corresponding sample data of each predetermined time step-length.
Wherein, financial transaction may include: stock exchange, futures exchange, foreign exchange, digital cash transaction etc..
Historical data refers to the data information of the financial transaction occurred within the past period, comprising: trading object, friendship
The easy amount of money, exchange hour etc..Wherein, trading object refers to the particular transactions product of financial transaction.For example, certain branch stock, certain number
Word currency etc..Wherein, half a year, 1 year, 2 years etc. be can be for a period of time, this is all within the scope of the present invention.
Since historical data amount is huge, if being all used to training pattern, a large amount of computing resources will be expended, therefore, it is necessary to
Sample data is selectively obtained, sample data training pattern is utilized.
When obtaining sample data, the sliding window of different predetermined time step-lengths can be set, utilize the sliding window
It slides in the historical data to obtain multiple window datas, each window data is sampled, each predetermined time is obtained
The corresponding sample data of step-length；If the historical data there are missing values, the position of the missing values in the historical data is obtained
And the digit of missing values, and filling of the value of the missing values maximum probability as missing values is calculated by Bayes's classification
Value.
Wherein, predetermined time step-length includes 6 chronomeres, 11 chronomeres and 16 chronomeres, chronomere
Refer to the granularity unit of historical data.For example, chronomere is day, minute etc..
For the sliding window of 6 chronomeres, the digit of corresponding window data is 6, the sample number sampled
According to digit be 6；For the sliding window of 11 chronomeres, the digit of corresponding window data is 11, and sampling obtains
The digit of sample data be 6, for example, the obtained sample data of sampling is (x1, x3, x5, x7, x9, x11), i.e. sample window
The the 1st, 3,5,7,9,11 data in mouth data；For the sliding window of 16 chronomeres, corresponding window data
Digit is 16, and the digit of the sample data sampled is 6, for example, the sample data that sampling obtains be (x1, x4, x7,
X10, x13, x16), i.e., the 1st, 4,7,10,13,16 data in sampling window data.
It is under limited computing resource by the purpose that the sliding window of different predetermined time step-lengths is arranged, expands institute
The degree remote and connection relationship of capturing information.Historical data without containing missing values is sampled to obtain sample data, is utilized
The sample data carrys out training pattern, can obtain the higher model of accuracy.
If there are missing values in historical data, the position of the missing values in the historical data and the position of missing values are obtained
Number, and Filling power of the value of the missing values maximum probability as missing values is calculated by Bayes's classification, to have supplemented
Whole historical data, and then ensure the integrality of sample data.
Filling power of the value as missing values that maximum probability is calculated by Bayes's classification is determined according to data attribute
The probability of each Filling power takes the Filling power of maximum probability to be filled by MapReduce.
Step S102: pre-processing the sample data, obtains training data.
When being pre-processed to sample data, comprising the following steps:
The first step cleans the sample data.
When cleaning data, according to the demand analysis of financial transaction, data category analysis, task definition, it is clear to obtain data
Wash scheme；Data are pre-processed, detect attribute error data, redundant data, and count to testing result；It determines dirty
The classification of data and corresponding cleaning program；According to conditional function, format function, summarize the constraint of analytic function definition integrity,
Inconsistent data reparation is carried out by integrity constraint；Utilize the clustering method automatic data-detection collection based on Euclidean distance
In attribute error, obtain modified data；Repeated data is cleared up by N-Gram algorithm；Clean data backflow is to sample number
According to library, data scrubbing cost can reduce by this method, improve the quality of data, guarantees the correctness and accuracy of data,
And then it can be improved the accuracy of prediction.
The sample data after cleaning is normalized in second step.
Since the data of financial trading market numerically differ larger with other influence factor data, and there is biggish wave
It is dynamic, therefore, it is necessary to which data are normalized, calculation formula are as follows:
Wherein, x is raw value, xminFor the smallest value of numerical value in all data of current dimension, xmaxFor current dimension institute
There are the maximum value of numerical value, X in data*For the normalization numerical value after scaling.
Third step classifies to the data after normalized, obtains training data；Wherein, training data is divided into instruction
Practice sample data, verify data and test data.
Wherein, three parts data proportion is followed successively by 70%, 20%, 10%, predicts mould for transaction value later
Training, verifying and the test of type.
When classifying to data, the data of each time window are categorized into three parts data all in accordance with aforementioned proportion
In, to guarantee the balance of data.
Step S103: utilizing the training data, is based on LSTM-RNN model, and training obtains transaction value prediction model.
Wherein, LSTM-RNN model indicates that shot and long term remembers Recognition with Recurrent Neural Network model.
Training data is being utilized, when training LSTM-RNN model, comprising the following steps:
The first step utilizes training sample data training LSTM-RNN model.
The present invention, which uses, has an input layer, 5 hidden layers, an output layer, and output layer uses identity function
Execute recurrence, hidden layer uses LSTM unit, for unit tool there are three door, input gate indicates whether the new dirt for allowing to acquire
Dye object concentration data information is added in currently hiding node layer, and if it is 1, door is opened, and allows to input, if it is 0, door
It closes, does not allow to input, can thus abandon some input information useless；Forget door to indicate whether to retain current hidden layer
The sample data of node storage, if it is 1, door opens reservation, and if it is 0, door is closed, and empties the sample that present node is stored
Data；Out gate, which indicates whether to export present node output valve, gives next layer (next hidden layer or output layer), if
It is 1, then door is opened, and the output valve of present node will act on next layer, and if it is 0, door is closed, and present node output valve is not defeated
Out.LSTM cellular construction compensates for the deficiency in traditional RNN structure, i.e., subsequent timing node perceives the timing node of front
Power decline.LSTM unit is a kind of special element for being referred to as memory cell, and be similar to accumulator and gate neuron: it is next
Time step will possess a weight and be connected to itself, copy the true value of oneself state and the external signal of accumulation, but this
Kind is by another modular learning and to determine that the multiplication gate for when removing memory content controls from connection, and particular content is as follows:
it=σ (Wxixt+Whiht-1+Wcict-1+bi)
ft=σ (Wxfxt+Whfht-1+Wcfct-1+bf)
ct=ftct-1+ittanh(Wxcxt+Whcht-1+bc)
ot=σ (Wxoxt+Whoht-1+Wcoct-1+bo)
ht=ottanh(ct)
Wherein σ is logic sigmoid function, and i, f, o, c, h respectively indicate input gate (input gate), forget door
(forget gate), out gate (output gate), unit activating vector (cell activation vectors) are hidden
Layer unit, Wxi、Whi、WciRespectively input feature value, hiding weight between layer unit, unit activating vector and input gate
Matrix, Wxf、Whf、WcfWeight square respectively between input feature value, hiding layer unit, unit activating vector and forgetting door
Battle array, Wxo、Who、WcoRespectively input feature value, hiding weight matrix between layer unit, unit activating vector and out gate,
Wxc、WhcWeight matrix respectively between input feature value, hiding layer unit and unit activating vector, the weight matrix
It is diagonal matrix；bi、bf、bc、boThe respectively deviation of input gate, forgetting door, out gate, unit activating vector, under t is used as
Sampling instant is indicated when mark, tanh is activation primitive.
Gate uses a sigmoid activation primitive, and input and cell state would generally use tanh activation primitive
To convert.When input is 0, the output of tanh function is 0.
In training pattern, using the type of attachment of dropout, i.e., the certain hidden layers of network are allowed immediately in model training
The weight of node does not work, idle node can temporarily not think be network structure a part, but its weight need to retain
Come (not updating temporarily only), to rework when the input of next sample.Network training process can be effectively prevented in dropout
In there is over-fitting.
RNN network structure training 10000epochs used in the present invention based on LSTM unit, learning rate
(learning rate) is 1, each epoch after training 2500epochs starts to reduce learning rate with coefficient 1.15.?
During trained each step, error vector is calculated according to cross entropy (cross entropy) criterion, according to standard backpropagation
Algorithm updates weight:
Error (t)=desired (t)-y (t)
Wherein desired is prediction output valve, and y (t) is real network output valve, and error is error amount.
Second step after having trained model using training sample data, needs to verify model using verify data, more
New parameter, detailed process are as follows:
By the model after verifying input training, verify data every iteration 1000 times progress during training network are primary
Test, final relatively test loss and train loss.When test loss is no longer reduced, network training is terminated, is indicated
RNN network verification comprising LSTM unit is completed.
Third step tests revised LSTM-RNN model using the test data；If being successfully tested, repair
LSTM-RNN model after just is transaction value prediction model；If test crash reacquires verify data, after training
LSTM-RNN model re-starts amendment.
When being tested using test data, test data is input in trained model, checks prediction data
With the gap between corresponding sample data, judge the gap whether within desired value, if, show to be successfully tested, can
To be predicted using the model；If not existing, show test crash, needs to be trained model again test, corrective networks
Parameter, and then step up precision of prediction.
Step S104: the following any time period is predicted using the transaction value prediction model according to current transaction data
And/or the data of financial transaction of any moment.
After establishing transaction value prediction model, according to current transaction data, following some moment or a certain is predicted
Data of financial transaction in a period after the completion of prediction, can be shown in a manner of visual, for example, column map grid
Formula, disk bitmap-format, chart format etc., this is all within the scope of the present invention.By using visual exhibition method,
It is able to use family and is more clearly seen that prediction result, improve user experience.
At each moment, financial trading market can all generate new data, all can serve as the data of training pattern, because
This needs to utilize newly generated current transaction data, the prediction of real-time optimization transaction value to guarantee the accuracy of model prediction
Model adjusts prototype network parameter.
In Optimized model, Adam algorithm can be used, Adam algorithm is one kind effectively based on the random optimization of gradient
Method, the algorithm fusion advantage of AdaGrad and RMSProp algorithm can calculate adaptability learning rate simultaneously to different parameters
And occupy less storage resource.Compared to other randomized optimization process, overall performance is more excellent in practical applications for Adam algorithm.
A kind of highly reliable model of LSTM model, the present invention by the prediction model based on LSTM-RNN model foundation,
Compared to the prior art, the data of financial transaction of the following any time period and/or any moment can be more accurately predicted, mention
High user experience.
More than, it is a kind of financial transaction price expectation method provided by the invention.
It is corresponding based on inventive concept identical with a kind of above-mentioned financial transaction price expectation method, the present invention
Embodiment additionally provides a kind of financial transaction price expectation device.Since Installation practice is substantially similar and embodiment of the method, institute
To describe fairly simple, the relevent part can refer to the partial explaination of embodiments of method.
A kind of financial transaction price expectation device provided by the invention, comprising:
Data acquisition module 101, for acquiring the historical data of financial transaction；The sliding of different predetermined time step-lengths is set
Window, using the sliding window the historical data slide to obtain multiple window datas, to each window data into
Row sampling, obtains the corresponding sample data of each predetermined time step-length；Data processing module 102, for the sample data into
Row pretreatment, obtains training data；
Training module 103 is based on LSTM-RNN model, it is pre- that training obtains transaction value for utilizing the training data
Survey model；
Prediction module 104, for according to current transaction data, using the transaction value prediction model, prediction is following to appoint
The data of financial transaction of one period and/or any moment.
In a specific embodiment provided by the invention, the data processing module 102, comprising:
Cleaning unit, for being cleaned to the sample data；
Normalization unit, for the sample data after cleaning to be normalized；
Taxon obtains training data for classifying to the data after normalized.
In a specific embodiment provided by the invention, the training data includes: training sample data, verify data
And test data, the training module 103, comprising:
Training unit, for utilizing training sample data training LSTM-RNN model；
Amending unit, for being modified using the verify data to the LSTM-RNN model after training；
Test cell, for being tested using the test data revised LSTM-RNN model；If test at
Function, then revised LSTM-RNN model is transaction value prediction model；If test crash, verify data is reacquired, it is right
LSTM-RNN model after training re-starts amendment.
In a specific embodiment provided by the invention, further includes:
Optimization module, for using Adam algorithm, using the current transaction data, transaction value described in real-time optimization is pre-
Survey model.
In a specific embodiment provided by the invention, the sample data of the financial transaction, comprising: financial product is handed over
Easy price, exchange hour and trading object.
In a specific embodiment provided by the invention, further includes:
Display module, for showing the following any time period and/or any moment predicted using visual means
Data of financial transaction.
More than, it is a kind of financial transaction price expectation device provided by the invention.
It is corresponding based on inventive concept identical with a kind of above-mentioned financial transaction price expectation method, the present invention
Embodiment additionally provides a kind of computer readable storage medium, is stored thereon with computer program, which is executed by processor
A kind of above-mentioned financial transaction price expectation method of Shi Shixian.
As shown from the above technical solution, a kind of computer readable storage medium provided in this embodiment, is stored thereon with meter
Calculation machine program when the program is executed by processor, can acquire the historical data of financial transaction；Sample is obtained from historical data
Data, then the sample data is pre-processed, obtain training data；Using the training data, it is based on LSTM-RNN mould
Type, training obtain transaction value prediction model；According to current transaction data, using the transaction value prediction model, prediction is not
Carry out the data of financial transaction of any time period and/or any moment.The present invention passes through the prediction based on LSTM-RNN model foundation
The financial transaction number of the following any time period and/or any moment can be more accurately predicted compared to the prior art in model
According to raising user experience.
It is corresponding based on inventive concept identical with a kind of above-mentioned financial transaction price expectation method, the present invention
Embodiment additionally provides a kind of financial transaction price expectation equipment, comprising: memory, processor and storage are on a memory and can
The computer program run on a processor, the processor realize that a kind of above-mentioned financial transaction price is pre- when executing described program
Survey method.
As shown from the above technical solution, a kind of financial transaction price expectation equipment provided in this embodiment, can acquire gold
Melt the historical data of transaction；Sample data is obtained from historical data, then the sample data is pre-processed, and is trained
Data；Using the training data, it is based on LSTM-RNN model, training obtains transaction value prediction model；According to current transaction
Data predict the data of financial transaction of the following any time period and/or any moment using the transaction value prediction model.
The present invention is by the prediction model based on LSTM-RNN model foundation, compared to the prior art, can be more accurately predicted not
Carry out the data of financial transaction of any time period and/or any moment, improves user experience.
In specification of the invention, numerous specific details are set forth.It is to be appreciated, however, that the embodiment of the present invention can be with
It practices without these specific details.In some instances, well known method, structure and skill is not been shown in detail
Art, so as not to obscure the understanding of this specification.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means specific features, structure, material or spy described in conjunction with this embodiment or example
Point is included at least one embodiment or example of the invention.In the present specification, schematic expression of the above terms are not
It must be directed to identical embodiment or example.Moreover, particular features, structures, materials, or characteristics described can be in office
It can be combined in any suitable manner in one or more embodiment or examples.In addition, without conflicting with each other, the skill of this field
Art personnel can tie the feature of different embodiments or examples described in this specification and different embodiments or examples
It closes and combines.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme should all cover within the scope of the claims and the description of the invention.
Claims (9)
1. a kind of financial transaction price expectation method characterized by comprising
Acquire the historical data of financial transaction；The sliding window of different predetermined time step-lengths is set, is existed using the sliding window
The historical data sliding samples each window data with obtaining multiple window datas, obtains each predetermined time step
Long corresponding sample data；
The sample data is pre-processed, training data is obtained；
Using the training data, it is based on LSTM-RNN model, training obtains transaction value prediction model；
The following any time period and/or any moment are predicted using the transaction value prediction model according to current transaction data
Data of financial transaction.
2. being instructed the method according to claim 1, wherein described pre-process the sample data
Practice data, comprising:
The sample data is cleaned；
Sample data after cleaning is normalized；
Classify to the data after normalized, obtains training data.
3. according to the method described in claim 2, it is characterized in that, the training data includes: training sample data, verifying number
According to and test data, it is described utilize the training data, be based on LSTM-RNN model, training obtain transaction value prediction model,
Include:
Utilize training sample data training LSTM-RNN model；
The LSTM-RNN model after training is modified using the verify data；
Revised LSTM-RNN model is tested using the test data；If being successfully tested, revised LSTM-
RNN model is transaction value prediction model；If test crash again classifies to sample data, verify data is obtained, it is right
LSTM-RNN model after training re-starts amendment.
4. method according to claim 1 or 3, which is characterized in that further include:
Using Adam algorithm, using the current transaction data, transaction value prediction model described in real-time optimization.
5. the method according to claim 1, wherein the sample data of the financial transaction, comprising: financial product
Transaction value, exchange hour and trading object.
6. the method according to claim 1, wherein further include:
The data of financial transaction of the following any time period and/or any moment predicted is shown using visual means.
7. a kind of financial transaction price expectation device characterized by comprising
Data acquisition module, for acquiring the historical data of financial transaction；The sliding window of different predetermined time step-lengths is set, benefit
It is slided with the sliding window in the historical data to obtain multiple window datas, each window data is sampled,
Obtain the corresponding sample data of each predetermined time step-length；
Data processing module obtains training data for pre-processing to the sample data；
Training module is based on LSTM-RNN model, training obtains transaction value prediction model for utilizing the training data；
Prediction module, for predicting the following any time period using the transaction value prediction model according to current transaction data
And/or the data of financial transaction of any moment.
8. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is held by processor
Method described in one of claim 1-6 is realized when row.
9. a kind of financial transaction price expectation equipment, comprising: memory, processor and storage are on a memory and can be in processor
The computer program of upper operation, which is characterized in that the processor is realized described in one of claim 1-6 when executing described program
Method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811442001.6A CN109636017A (en) | 2018-11-29 | 2018-11-29 | A kind of financial transaction price expectation method, apparatus, medium and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811442001.6A CN109636017A (en) | 2018-11-29 | 2018-11-29 | A kind of financial transaction price expectation method, apparatus, medium and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109636017A true CN109636017A (en) | 2019-04-16 |
Family
ID=66069688
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811442001.6A Pending CN109636017A (en) | 2018-11-29 | 2018-11-29 | A kind of financial transaction price expectation method, apparatus, medium and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109636017A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569190A (en) * | 2019-08-27 | 2019-12-13 | 中国工商银行股份有限公司 | Transaction pressure testing method and device, electronic device and readable storage medium |
| CN111415270A (en) * | 2020-03-03 | 2020-07-14 | 浙江万胜智能科技股份有限公司 | Power load intelligent identification method based on L STM learning |
| CN111709532A (en) * | 2020-05-26 | 2020-09-25 | 重庆大学 | Model-independent local interpretation-based online shopping representative sample selection system |
| CN111798263A (en) * | 2020-05-22 | 2020-10-20 | 北京国电通网络技术有限公司 | Transaction trend prediction method and device |
| CN112070535A (en) * | 2020-09-03 | 2020-12-11 | 常州微亿智造科技有限公司 | Electric vehicle price prediction method and device |
| CN113781219A (en) * | 2021-09-06 | 2021-12-10 | 上海卡方信息科技有限公司 | Real-time algorithm trading system and method in stock trading process |
| CN115545790A (en) * | 2022-10-20 | 2022-12-30 | 北京宽客进化科技有限公司 | Price data prediction method and device, electronic equipment and storage medium |
-
2018
- 2018-11-29 CN CN201811442001.6A patent/CN109636017A/en active Pending
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569190A (en) * | 2019-08-27 | 2019-12-13 | 中国工商银行股份有限公司 | Transaction pressure testing method and device, electronic device and readable storage medium |
| CN110569190B (en) * | 2019-08-27 | 2023-03-31 | 中国工商银行股份有限公司 | Transaction pressure testing method and device, electronic device and readable storage medium |
| CN111415270A (en) * | 2020-03-03 | 2020-07-14 | 浙江万胜智能科技股份有限公司 | Power load intelligent identification method based on L STM learning |
| CN111798263A (en) * | 2020-05-22 | 2020-10-20 | 北京国电通网络技术有限公司 | Transaction trend prediction method and device |
| CN111709532A (en) * | 2020-05-26 | 2020-09-25 | 重庆大学 | Model-independent local interpretation-based online shopping representative sample selection system |
| CN111709532B (en) * | 2020-05-26 | 2023-09-22 | 重庆大学 | Online shopping representative sample selection system based on model-independent local interpretation |
| CN112070535A (en) * | 2020-09-03 | 2020-12-11 | 常州微亿智造科技有限公司 | Electric vehicle price prediction method and device |
| CN113781219A (en) * | 2021-09-06 | 2021-12-10 | 上海卡方信息科技有限公司 | Real-time algorithm trading system and method in stock trading process |
| CN115545790A (en) * | 2022-10-20 | 2022-12-30 | 北京宽客进化科技有限公司 | Price data prediction method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109636017A (en) | A kind of financial transaction price expectation method, apparatus, medium and equipment | |
| CN109255506B (en) | Internet financial user loan overdue prediction method based on big data | |
| Chiang et al. | An adaptive stock index trading decision support system | |
| Safari et al. | International stock index prediction using artificial neural network (ANN) and Python programming | |
| Skuratov et al. | Creation of a neural network algorithm for automated collection and analysis of statistics of exchange quotes graphics | |
| Kroha et al. | Hurst Exponent and Trading Signals Derived from Market Time Series. | |
| Li et al. | Financial risk prediction for listed companies using IPSO-BP neural network | |
| Li et al. | Stock trading strategies based on deep reinforcement learning | |
| Song et al. | Incorporating research reports and market sentiment for stock excess return prediction: a case of mainland china | |
| Jabeen et al. | An Empirical Study of Macroeconomic Factors and Stock Returns in the Context of Economic Uncertainty News Sentiment Using Machine Learning | |
| CN110458719A (en) | A kind of dispatching method of electric network and system for power grid enterprises | |
| Ghule et al. | Gold price prediction using machine learning | |
| Rofik et al. | The Optimization of Credit Scoring Model Using Stacking Ensemble Learning and Oversampling Techniques | |
| CN109919366A (en) | Forecasting of Stock Prices method based on tensor and event-driven LSTM model | |
| Heryadi et al. | Foreign exchange prediction using machine learning approach: A pilot study | |
| Feng | Bank Green Credit Risk Assessment and Management by Mobile Computing and Machine Learning Neural Network under the Efficient Wireless Communication | |
| Dhuhita et al. | Gold Price Prediction Based On Yahoo Finance Data Using Lstm Algorithm | |
| Talebi et al. | Evidence for the Ability of the Regression Model and Particle Swarm Optimization Algorithm in Predicting Future Cash Flows | |
| KR102420952B1 (en) | Loan expansion hypothesis testing system using artificial intelligence and method using the same | |
| BOLDEA et al. | THE FINANCIAL CRISIS PREDICTION USING AN ARTIFICIAL INTELLIGENCE METHOD. | |
| Ertuna | Stock Market Prediction Using Neural Network and Time Series Forecasting | |
| US20230252387A1 (en) | Apparatus, method and recording medium storing commands for providing artificial-intelligence-based risk management solution in credit exposure business of financial institution | |
| Kavitha et al. | Stock Closing Price Prediction Using Deep Learning, TensorFlow, and Keras | |
| Mustafa et al. | Stock market prediction with the integration of daily news sentiments using Long Short Term Memory in Artificial Neural Network with the ensemble approach | |
| Goswami | Predicting the Stock Price/Movements of Companies Listed Under Bse-It (Bombay Stock Exchange) |
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 |
Application publication date: 20190416 |
|
| RJ01 | Rejection of invention patent application after publication |