CN109615449A - A kind of prediction technique and device, a kind of calculating equipment and storage medium - Google Patents
A kind of prediction technique and device, a kind of calculating equipment and storage medium Download PDFInfo
- Publication number
- CN109615449A CN109615449A CN201811248271.3A CN201811248271A CN109615449A CN 109615449 A CN109615449 A CN 109615449A CN 201811248271 A CN201811248271 A CN 201811248271A CN 109615449 A CN109615449 A CN 109615449A
- Authority
- CN
- China
- Prior art keywords
- moment
- lstm
- data
- prediction
- hidden layer
- 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
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0609—Buyer or seller confidence or verification
-
- 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
- G06Q30/00—Commerce
- G06Q30/01—Customer relationship services
- G06Q30/012—Providing warranty services
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Software Systems (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)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of prediction technique and device, a kind of calculating equipment and storage medium that this specification provides, wherein, the method includes receiving the input data at t-1 moment, wherein, the input data includes history achievement data, periodic data and influence factor data, and the t is current time；Select the shot and long term memory network LSTM hidden layer at t-2 moment；According to the LSTM result of the t moment of the LSTM hidden layer at the t-2 moment, the input data at the t-1 moment and the LSTM model output prediction.
Description
Technical field
This application involves field of computer technology, in particular to a kind of prediction technique and device, a kind of calculating equipment and deposit
Storage media.
Background technique
Ten hundreds of with the trade company of Alipay cooperation, the risk indicator (such as returning commission volume, tip-offs about environmental issues etc.) of trade company is from wind
It is particularly significant from the point of view of controlling in angle, such as find that risk indicator there are unusual fluctuations, then needs to pay close attention to trade company.Such as trade company/clothes
It returned commission volume on the day of business quotient and uprushed, trade company's turnover is uprushed/anticlimax, and trade company's tip-offs about environmental issues (complaint for having not enough time to trial) are than flat
Day increases, then needs to pay close attention to trade company, extract investigate to trade company's situation in time.But too due to trade company's quantity
Greatly, manually see that data exception is checked one by one, screen out has abnormal trade company compared with passing, and speed is slow and workload
Greatly, working efficiency is low.
Summary of the invention
In view of this, the embodiment of the present application provides a kind of prediction technique and device, a kind of calculating equipment and storage medium,
To solve technological deficiency existing in the prior art.
In a first aspect, this specification embodiment discloses a kind of prediction technique, comprising:
Receive the input data at t-1 moment, wherein the input data includes history achievement data, periodic data
And influence factor data, and the t is current time；
Select the shot and long term memory network LSTM hidden layer at t-2 moment；
According to the LSTM hidden layer at the t-2 moment, the input data at the t-1 moment and the LSTM model
Export the LSTM result of the t moment of prediction.
Second aspect, this specification embodiment also disclose a kind of prediction meanss, comprising:
Receiving module is configured as receiving the input data at t-1 moment, wherein the input data includes that history refers to
Data, periodic data and influence factor data are marked, and the t is current time；
Selecting module is configured as the shot and long term memory network LSTM hidden layer at selection t-2 moment；
Prediction module is configured as the LSTM hidden layer according to the t-2 moment, the input number at the t-1 moment
Accordingly and the LSTM model output prediction t moment LSTM result.
The third aspect, this specification embodiment also disclose a kind of calculating equipment, including memory, processor and are stored in
On memory and the computer instruction that can run on a processor, the processor realize that the instruction is located when executing described instruction
The step of reason device realizes prediction technique as described above when executing.
Fourth aspect, this specification embodiment also disclose a kind of computer readable storage medium, are stored with computer
The step of instruction, which realizes prediction technique as described above when being executed by processor.
A kind of prediction technique and device, a kind of calculating equipment and storage medium that this specification provides, wherein the method
Including receive the t-1 moment input data, wherein the input data include history achievement data, periodic data and
Influence factor data, and the t is current time；Select the shot and long term memory network LSTM hidden layer at t-2 moment；According to
The LSTM hidden layer at the t-2 moment, the input data at the t-1 moment and LSTM model output prediction the
The LSTM result of t moment.
Detailed description of the invention
Fig. 1 is a kind of flow chart for prediction technique that this specification one or more embodiment provides；
Fig. 2 is a kind of flow chart for prediction technique that this specification one or more embodiment provides；
Fig. 3 is a kind of flow chart for prediction technique that this specification one or more embodiment provides；
Fig. 4-1 is that a kind of LSTM model for prediction technique that this specification one or more embodiment provides forgets door
Schematic diagram；
Fig. 4-2 is a kind of LSTM mode input door for prediction technique that this specification one or more embodiment provides
Schematic diagram；
Fig. 4-3 is a kind of LSTM mode input door for prediction technique that this specification one or more embodiment provides
Schematic diagram；
Fig. 4-4 is a kind of LSTM model output door for prediction technique that this specification one or more embodiment provides
Schematic diagram；
Fig. 5 is a kind of structural schematic diagram for prediction meanss that this specification one or more embodiment provides；
Fig. 6 is a kind of block schematic illustration for calculating equipment that this specification one or more embodiment provides.
Specific embodiment
Many details are explained in the following description in order to fully understand the application.But the application can be with
Much it is different from other way described herein to implement, those skilled in the art can be without prejudice to the application intension the case where
Under do similar popularization, therefore the application is not limited by following public specific implementation.
The term used in this specification one or more embodiment be only merely for for the purpose of describing particular embodiments,
It is not intended to be limiting this specification one or more embodiment.In this specification one or more embodiment and appended claims
The "an" of singular used in book, " described " and "the" are also intended to including most forms, unless context is clearly
Indicate other meanings.It is also understood that term "and/or" used in this specification one or more embodiment refers to and includes
One or more associated any or all of project listed may combine.
It will be appreciated that though may be retouched using term first, second etc. in this specification one or more embodiment
Various information are stated, but these information should not necessarily be limited by these terms.These terms are only used to for same type of information being distinguished from each other
It opens.For example, first can also be referred to as second, class in the case where not departing from this specification one or more scope of embodiments
As, second can also be referred to as first.Depending on context, word as used in this " if " can be construed to
" ... when " or " when ... " or " in response to determination ".
Firstly, the vocabulary of terms being related to one or more embodiments of the invention explains.
LSTM: shot and long term memory network or time recurrent neural network, commonly used in behavior sequence, natural language
Translation etc., can solve long sequence Dependence Problem.
Referring to Fig. 1, this specification one or more embodiment provides a kind of flow chart of prediction technique, including input layer
102, hidden layer 104 and output layer 106.
Wherein, the input of input layer 102 is variable X 1, X2 ... the Xn of LSTM model, each input layer 102
Variable X 1, X2 ... Xn can be identical, can not also be identical, and variable not necessarily needs to be significantly strong variable, because
Low weight can be assigned to the variable on relevant property ground by self study for LSTM model, it can be to the change on no relevance ground
It measures and assigns high weight, in practical application, variable setting can be carried out according to actual needs, the application does not make any limit to this
It is fixed.
For example, t-n days variables of input layer 102 may include the trading volume of the t-n days each trade companies, the t-n days
All trade companies trading volume, what day is within the t-n days, be New Year's Day within the t-n days；T-n+1 days variables of input layer 102 can be with
The trading volume of trading volume, the t-n+1 days all trade companies including the t-n+1 days each trade companies is week in the t-n+1 days
It is several, have advertising campaign within t-n+1 days；T-1 days variables of input layer 102 may include the transaction of the t-1 days each trade companies
Amount, the trading volume of the t-1 days all trade companies, what day is within the t-1 days, the has advertising campaign etc. for t-1 days.
After variable X 1, X2 ... the Xn for the LSTM model that the input layer 102 inputs enters LSTM model, it can enter described hidden
Containing layer 104.
The hidden layer 104 can be inputted to the same day the t-1 days variables and the hidden layer 104 of the t-2 days upper one days
Data are processed together, and the variable that the t-2 days hidden layers contain the t-2 days simultaneously inputs and the t-3 days hidden layers
Data, and so on, the hidden layer of every day contains the data of the variable input and the hidden layer of upper one day on the same day simultaneously,
So that the LSTM model carries out not only considering current input when prediction of result, it is also contemplated that history input before, so that prediction
As a result more accurate.
What the output layer 106 exported be need to predict as a result, such as the t-0 days trading volume, complaint amount.
It in this specification one or more embodiment, is given a forecast using LSTM model, can preferably combine history index
Trade company's trading volume of data such as yesterdays or transaction amount etc., extraneous factor are for example whether have advertising campaign etc., data week
Phase property feature is for example whether be weekend etc., more accurately results such as the trading volume or complaint amount on the prediction same day.
Referring to fig. 2, this specification one or more embodiment provides a kind of prediction technique, including step 202 is to step
206。
Step 202: receiving the input data at t-1 moment, wherein the input data includes history achievement data, week
Phase property data and influence factor data, and the t is current time.
In this specification one or more embodiment, the moment as unit of day, using week as the period, the periodicity
Data include using week as certain day in the period, i.e. Monday, Tuesday, Wednesday, Thursday, Friday, Saturday or Sunday.
The t is that the current time i.e. described t is the same day, and the t-1 moment is upper one day, and the t-2 moment is then institute
State upper one day at t-1 moment.
In this specification one or more embodiment, the history achievement data includes the trading volume and transaction of certain trade company
The trading volume and transaction amount of the amount of money or all trade companies.
The influence factor data include whether as festivals or holidays or whether have advertising campaign.
In this specification one or more embodiment, if the input data includes history achievement data, periodic data
And influence factor data, and the t is current time, then the input data for receiving the t-1 moment may include the t-1 days
The trading volume of certain trade company, has advertising campaign etc. for t-1 days for weekend and the in the t-1 days at the trading volume of the t-1 days all trade companies.
Step 204: the shot and long term memory network LSTM hidden layer at selection t-2 moment.
In this specification one or more embodiment, the shot and long term memory network LSTM hidden layer packet at the t-2 moment
The t-2 days input datas and the t-3 days hidden layer d are contained.
Step 206: according to the LSTM hidden layer at the t-2 moment, the input data at the t-1 moment and described
The LSTM result of the t moment of LSTM model output prediction.
In this specification one or more embodiment, according to the t-1 days input datas, the selection the t-2 days
LSTM hidden layer by the LSTM model output prediction the t days LSTM as a result, such as trading volume, transaction amount or
It is complaint amount etc..
In this specification one or more embodiment, period sexual factor can be added in the LSTM model, keep prediction more preferably quasi-
Really, and the LSTM model can not only learn itself history achievement data, and more variables can also be added (as outside
Whether boundary's factor is the Spring Festival or whether has advertising campaign etc.), LSTM model prediction can be allowed more preferably accurate.
Referring to Fig. 3, this specification one or more embodiment provides a kind of prediction technique, including step 302 is to step
308。
Step 302: receiving the input data at t-1 moment, wherein the input data includes history achievement data, week
Phase property data and influence factor data, and the t is current time.
Step 304: the shot and long term memory network LSTM hidden layer at selection t-2 moment.
Step 306: according to the LSTM hidden layer at the t-2 moment, the input data at the t-1 moment and described
The LSTM result of the t moment of LSTM model output prediction.
Step 308: if the difference of the LSTM result of the t moment of prediction and actual result is greater than preset threshold, issuing pre-
Alert prompt.
In this specification one or more embodiment, for the different time (t-1, t-2 or t-n), what they entered
Hidden layer be it is the same, parameter is all unified.Model compares lightweight in this way.But hidden layer has " memory " can be it
The variable input of preceding data selectively preserves, and predicts the output on the same day together with the variable input on the same day.
In this specification one or more embodiment, the LSTM model can pass through the data characteristics of main body, self study tune
Whole weight (such as adapting to periodic data to periodical weighted value, subtract weight to adapt to aperiodicity weight), can be preferably
In conjunction with the data of metric history, extraneous factor, data periodic feature etc. more accurately predicts certain data on the same day；So
It is compared afterwards according to the data of prediction with real data, it is capable of emitting pre- if the data and real data difference of prediction are larger
Alert prompt, personnel convenient for safeguarding can check exception.
The accuracy rate that prediction result can be improved using the LSTM model of this specification one or more embodiment, into
Row early warning, which is checked, can reduce false alarm degree when intensity of anomaly, reduce the rate that reports an error.
- 1 to Fig. 4-4 referring to fig. 4, this specification one or more embodiment provide a kind of prediction technique, the prediction
The LSTM model in method includes input gate, forgets door, out gate, state cell and LSTM as a result, respectively by following
Formula, which calculates, to be obtained:
it=σ (Wi.[ht-1, xt]+bi)
ft=σ (Wf.[ht-1, xt]+bf)
Ot=σ (Wo.[ht-1, xt]+bo)
ht=Ot*tanh(Ct)
Wherein, xtFor the input data at the t-1 moment, the ht-1For the LSTM hidden layer at t-2 moment, W is power
Weight matrix, b are bias vector, and i, f, o, c, h are respectively the input gate, forget door, out gate, state cell and LSTM knot
Fruit, σ are sigmoid function, the calculation formula of the sigmoid function are as follows:The calculating of the tanh function
Formula are as follows:
There is the mechanism of the forgetting of similar information selectivity in the LSTM model, i.e., forgetting door, the forgetting door can bases
The feature of training data, study or forgetting (discarding) of the self study to past Input variable selection.For example, when there is data
When having periodical, the LSTM model, which can choose, increases weight to the farther time, but reduces weight to the closer time,
The accuracy rate of model prediction is improved, error is reduced.Such self study system helps to adapt to different main body/trade companies, data
Characteristic it is different.Periodicity and aperiodicity all can adjust the weight of hidden layer in LSTM model, study in self study
One is adapted to the LSTM prediction model of different subjects.
For example, LSTM model according to 30 days or 60 days historical datas, learns to the sales volume at each weekend compare
Sales volume in week is risen, then the same day if desired predicted is determined as weekend in practical application, then can increase power
Weight, so that the predicted value on the same day obtained using LSTM model is more close with the value at history weekend；If desired the same day predicted is not
It is weekend, then can reduce weight, so that the predicted value on the same day obtained using LSTM model is more close with the value in history week.
In this specification one or more embodiment, the LSTM model can be by main body/trade company data characteristics, to week
Phase property weighted value adapts to periodic data, subtracts weight to adapt to aperiodicity weight, so that prediction result is more accurate, so that
It is lower in progress early warning and alert Times error rate to obtain the LSTM model.
Referring to Fig. 5, this specification one or more embodiment provides a kind of prediction meanss, comprising:
Receiving module 502 is configured as receiving the input data at t-1 moment, wherein the input data includes history
Achievement data, periodic data and influence factor data, and the t is current time；
Selecting module 504 is configured as the shot and long term memory network LSTM hidden layer at selection t-2 moment；
Prediction module 506 is configured as the LSTM hidden layer according to the t-2 moment, the input at the t-1 moment
The LSTM result of data and the t moment of LSTM model output prediction.
Optionally, the history achievement data includes the trading volume of certain trade company and the friendship of transaction amount or all trade companies
Easily amount and transaction amount.
Optionally, the moment, using week as the period, the periodic data included using week as the period as unit of day
In certain day.
Optionally, the influence factor data include whether as festivals or holidays or whether have advertising campaign.
Optionally, described device further include:
If the LSTM result of the t moment of prediction and the difference of actual result are greater than preset threshold, early warning is issued.
Optionally, the LSTM model includes input gate, forgets door, out gate, state cell and LSTM as a result, leading to respectively
It crosses following formula and calculates acquisition:
it=σ (Wi.[ht-1, xt]+bi)
ft=σ (Wf.[ht-1, xt]+bf)
Ot=σ (Wo.[ht-1, xt]+bo)
ht=Ot*tanh(Ct)
Wherein, xtFor the input data at the t-1 moment, the ht-1For the LSTM hidden layer at t-2 moment, W is power
Weight matrix, b are bias vector, and i, f, o, c, h are respectively the input gate, forget door, out gate, state cell and LSTM knot
Fruit, σ are sigmoid function, the calculation formula of the sigmoid function are as follows:The calculating of the tanh function
Formula are as follows:
In this specification one or more embodiment, the LSTM model can pass through the data characteristics of main body, self study tune
Whole weight (such as adapting to periodic data to periodical weighted value, subtract weight to adapt to aperiodicity weight), can be preferably
In conjunction with the data of metric history, extraneous factor, data periodic feature etc. more accurately predicts certain data on the same day；So
It is compared afterwards according to the data of prediction with real data, it is capable of emitting pre- if the data and real data difference of prediction are larger
Alert prompt, personnel convenient for safeguarding can check exception.
Fig. 6 is to show the structural block diagram of the calculating equipment 600 according to one embodiment of this specification.The calculating equipment 600
Component include but is not limited to memory 610 and processor 620.Processor 620 is connected with memory 610 by bus 630,
Database 650 is for saving data.
Calculating equipment 600 further includes access device 640, access device 640 enable calculate equipment 600 via one or
Multiple networks 660 communicate.The example of these networks includes public switched telephone network (PSTN), local area network (LAN), wide area network
(WAN), the combination of the communication network of personal area network (PAN) or such as internet.Access device 640 may include wired or wireless
One or more of any kind of network interface (for example, network interface card (NIC)), such as IEEE802.11 wireless local area
Net (WLAN) wireless interface, worldwide interoperability for microwave accesses (Wi-MAX) interface, Ethernet interface, universal serial bus (USB) connect
Mouth, cellular network interface, blue tooth interface, near-field communication (NFC) interface, etc..
In one embodiment of this specification, unshowned other component in above-mentioned and Fig. 6 of equipment 600 is calculated
It can be connected to each other, such as pass through bus.It should be appreciated that calculating device structure block diagram shown in fig. 6 is merely for the sake of example
Purpose, rather than the limitation to this specification range.Those skilled in the art can according to need, and increase or replace other portions
Part.
Calculating equipment 600 can be any kind of static or mobile computing device, including mobile computer or mobile meter
Calculate equipment (for example, tablet computer, personal digital assistant, laptop computer, notebook computer, net book etc.), movement
Phone (for example, smart phone), wearable calculating equipment (for example, smartwatch, intelligent glasses etc.) or other kinds of shifting
Dynamic equipment, or the static calculating equipment of such as desktop computer or PC.Calculating equipment 600 can also be mobile or state type
Server.
One embodiment of the application also provides a kind of computer readable storage medium, is stored with computer instruction, the instruction
The step of prediction technique as previously described is realized when being executed by processor.
A kind of exemplary scheme of above-mentioned computer readable storage medium for the present embodiment.It should be noted that this is deposited
The technical solution of the technical solution of storage media and above-mentioned prediction technique belongs to same design, and the technical solution of storage medium is not detailed
The detail content carefully described may refer to the description of the technical solution of above-mentioned prediction technique.
The computer instruction includes computer program code, the computer program code can for source code form,
Object identification code form, executable file or certain intermediate forms etc..The computer-readable medium may include: that can carry institute
State any entity or device, recording medium, USB flash disk, mobile hard disk, magnetic disk, CD, the computer storage of computer program code
Device, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory),
Electric carrier signal, telecommunication signal and software distribution medium etc..It should be noted that the computer-readable medium include it is interior
Increase and decrease appropriate can be carried out according to the requirement made laws in jurisdiction with patent practice by holding, such as in certain jurisdictions of courts
Area does not include electric carrier signal and telecommunication signal according to legislation and patent practice, computer-readable medium.
It should be noted that for the various method embodiments described above, describing for simplicity, therefore, it is stated as a series of
Combination of actions, but those skilled in the art should understand that, the application is not limited by the described action sequence because
According to the application, certain steps can use other sequences or carry out simultaneously.Secondly, those skilled in the art should also know
It knows, the embodiments described in the specification are all preferred embodiments, and related actions and modules might not all be this Shen
It please be necessary.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, it may refer to the associated description of other embodiments.
The application preferred embodiment disclosed above is only intended to help to illustrate the application.There is no detailed for alternative embodiment
All details are described, are not limited the invention to the specific embodiments described.Obviously, according to the content of this specification,
It can make many modifications and variations.These embodiments are chosen and specifically described to this specification, is in order to preferably explain the application
Principle and practical application, so that skilled artisan be enable to better understand and utilize the application.The application is only
It is limited by claims and its full scope and equivalent.
Claims (10)
1. a kind of prediction technique characterized by comprising
Receive the input data at t-1 moment, wherein the input data include history achievement data, periodic data and
Influence factor data, and the t is current time；
Select the shot and long term memory network LSTM hidden layer at t-2 moment；
It is exported according to the LSTM hidden layer at the t-2 moment, the input data at the t-1 moment and the LSTM model
The LSTM result of the t moment of prediction.
2. the method according to claim 1, wherein the history achievement data include the trading volume of certain trade company with
And the trading volume and transaction amount of transaction amount or all trade companies.
3. the method according to claim 1, wherein the moment as unit of day, it is described using week as the period
Periodic data includes using week as certain day in the period.
4. the method according to claim 1, wherein the influence factor data include whether as festivals or holidays or
Whether advertising campaign is had.
5. the method according to claim 1, wherein further include:
If the LSTM result of the t moment of prediction and the difference of actual result are greater than preset threshold, early warning is issued.
6. the method according to claim 1, wherein the LSTM model includes input gate, forgets door, output
Door, state cell and LSTM as a result, be calculated by the following formula acquisition respectively:
it=σ (Wi.[ht-1, xt]+bi)
ft=σ (Wf.[ht-1, xt]+bf)
Ot=σ (Wo.[ht-1, xt]+bo)
ht=Ot*tanh(Ct)
Wherein, xtFor the input data at the t-1 moment, the ht-1For the LSTM hidden layer at t-2 moment, W is weight square
Battle array, b are bias vector, and i, f, o, c, h are respectively the input gate, forget door, out gate, state cell and LSTM as a result, σ is
Sigmoid function, the calculation formula of the sigmoid function are as follows:The calculation formula of the tanh function
Are as follows:
7. a kind of prediction meanss characterized by comprising
Receiving module is configured as receiving the input data at t-1 moment, wherein the input data includes history index number
According to, periodic data and influence factor data, and the t is current time；
Selecting module is configured as the shot and long term memory network LSTM hidden layer at selection t-2 moment；
Prediction module, be configured as the LSTM hidden layer according to the t-2 moment, the input data at the t-1 moment with
And the LSTM result of the t moment of the LSTM model output prediction.
8. device according to claim 7, which is characterized in that the history achievement data include the trading volume of certain trade company with
And the trading volume and transaction amount of transaction amount or all trade companies.
9. a kind of calculating equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine instruction, which is characterized in that the processor is realized when executing described instruction realizes that right the is wanted when instruction is executed by processor
The step of seeking 1-6 any one the method.
10. a kind of computer readable storage medium, is stored with computer instruction, which is characterized in that the instruction is held by processor
The step of claim 1-6 any one the method is realized when row.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811248271.3A CN109615449A (en) | 2018-10-25 | 2018-10-25 | A kind of prediction technique and device, a kind of calculating equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811248271.3A CN109615449A (en) | 2018-10-25 | 2018-10-25 | A kind of prediction technique and device, a kind of calculating equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109615449A true CN109615449A (en) | 2019-04-12 |
Family
ID=66002857
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811248271.3A Pending CN109615449A (en) | 2018-10-25 | 2018-10-25 | A kind of prediction technique and device, a kind of calculating equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109615449A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110347506A (en) * | 2019-06-28 | 2019-10-18 | Oppo广东移动通信有限公司 | Data processing method, device, storage medium and electronic equipment based on LSTM |
| CN110610382A (en) * | 2019-09-10 | 2019-12-24 | 浙江大搜车软件技术有限公司 | Vehicle sales prediction method, apparatus, computer device, and storage medium |
| CN115632887A (en) * | 2022-12-21 | 2023-01-20 | 国网数字科技控股有限公司 | Block chain network abnormal data detection method, device and equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150356075A1 (en) * | 2014-06-06 | 2015-12-10 | Google Inc. | Generating representations of input sequences using neural networks |
| CN105513591A (en) * | 2015-12-21 | 2016-04-20 | 百度在线网络技术（北京）有限公司 | Method and device for speech recognition by use of LSTM recurrent neural network model |
| CN106952181A (en) * | 2017-03-08 | 2017-07-14 | 深圳市景程信息科技有限公司 | Electric Load Prediction System based on long Memory Neural Networks in short-term |
| CN107992957A (en) * | 2016-10-25 | 2018-05-04 | 航天信息股份有限公司 | A kind of method and device for prediction of being carried on the work based on tax data |
| CN108510065A (en) * | 2018-03-30 | 2018-09-07 | 中国科学院计算技术研究所 | Computing device and computational methods applied to long Memory Neural Networks in short-term |
| CN108550047A (en) * | 2018-03-20 | 2018-09-18 | 阿里巴巴集团控股有限公司 | The prediction technique and device of trading volume |
| CN108596387A (en) * | 2018-04-20 | 2018-09-28 | 阿里巴巴集团控股有限公司 | A kind of prediction technique of fund, device and electronic equipment |
-
2018
- 2018-10-25 CN CN201811248271.3A patent/CN109615449A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150356075A1 (en) * | 2014-06-06 | 2015-12-10 | Google Inc. | Generating representations of input sequences using neural networks |
| CN105513591A (en) * | 2015-12-21 | 2016-04-20 | 百度在线网络技术（北京）有限公司 | Method and device for speech recognition by use of LSTM recurrent neural network model |
| CN107992957A (en) * | 2016-10-25 | 2018-05-04 | 航天信息股份有限公司 | A kind of method and device for prediction of being carried on the work based on tax data |
| CN106952181A (en) * | 2017-03-08 | 2017-07-14 | 深圳市景程信息科技有限公司 | Electric Load Prediction System based on long Memory Neural Networks in short-term |
| CN108550047A (en) * | 2018-03-20 | 2018-09-18 | 阿里巴巴集团控股有限公司 | The prediction technique and device of trading volume |
| CN108510065A (en) * | 2018-03-30 | 2018-09-07 | 中国科学院计算技术研究所 | Computing device and computational methods applied to long Memory Neural Networks in short-term |
| CN108596387A (en) * | 2018-04-20 | 2018-09-28 | 阿里巴巴集团控股有限公司 | A kind of prediction technique of fund, device and electronic equipment |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110347506A (en) * | 2019-06-28 | 2019-10-18 | Oppo广东移动通信有限公司 | Data processing method, device, storage medium and electronic equipment based on LSTM |
| CN110347506B (en) * | 2019-06-28 | 2023-01-06 | Oppo广东移动通信有限公司 | Data processing method and device based on LSTM, storage medium and electronic equipment |
| CN110610382A (en) * | 2019-09-10 | 2019-12-24 | 浙江大搜车软件技术有限公司 | Vehicle sales prediction method, apparatus, computer device, and storage medium |
| CN115632887A (en) * | 2022-12-21 | 2023-01-20 | 国网数字科技控股有限公司 | Block chain network abnormal data detection method, device and equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Khan et al. | Machine learning based hybrid system for imputation and efficient energy demand forecasting | |
| Blattberg et al. | Database models and managerial intuition: 50% model+ 50% manager | |
| McAleer et al. | A decision rule to minimize daily capital charges in forecasting value‐at‐risk | |
| CN109615449A (en) | A kind of prediction technique and device, a kind of calculating equipment and storage medium | |
| Lai et al. | Load forecasting based on deep neural network and historical data augmentation | |
| CN103002164A (en) | Telephone traffic forecasting method of electric power call center | |
| Cipiloglu Yildiz et al. | A portfolio construction framework using LSTM‐based stock markets forecasting | |
| Malikov et al. | Estimation of firm‐level productivity in the presence of exports: Evidence from China's manufacturing | |
| Houssein et al. | Assess deep learning models for Egyptian exchange prediction using nonlinear artificial neural networks | |
| Liu et al. | LSTM based reserve prediction for bank outlets | |
| CN109615414A (en) | House property predictor method, device and storage medium | |
| Jain et al. | Prediction and analysis of financial trends using Ml algorithm | |
| Singha et al. | Application of different machine learning models for supply chain demand forecasting: comparative analysis | |
| Chen et al. | Stochastic configuration network based on improved whale optimization algorithm for nonstationary time series prediction | |
| Xinliang et al. | University restaurant sales forecast based on bp neural network–in shanghai jiao tong university case | |
| Yan et al. | An Experimental Machine Learning Approach for Mid-Term Energy Demand Forecasting | |
| Wan et al. | Deep learning-based management for wastewater treatment plants under blockchain environment | |
| CN114781937A (en) | Method and device for pre-paid card enterprise risk early warning and storage medium | |
| US11847144B2 (en) | Systems and methods of correlating database entries for automated metric computation | |
| Holderbaum et al. | Short Term Load Forecasting (STLF) | |
| CN112116253A (en) | Method, device and system for selecting central mesh point | |
| Steens et al. | A forecast-based load management approach for commercial buildings demonstrated on an integration of BEV | |
| CN117829375B (en) | Method, device, equipment and medium for predicting multi-region demand of inter-city shuttle passenger transport | |
| Shabri et al. | Forecasting Tourist Arrivals Using a Combination of Long Short-Term Memory and Fourier Series | |
| Yin et al. | Sliding Window GBDT for Electricity Demand Forecasting |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200918 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Applicant before: Alibaba Group Holding Ltd. Effective date of registration: 20200918 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant after: Innovative advanced technology Co.,Ltd. Address before: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant before: Advanced innovation technology Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190412 |
|
| RJ01 | Rejection of invention patent application after publication |