CN108550047A - The prediction technique and device of trading volume - Google Patents
The prediction technique and device of trading volume Download PDFInfo
- Publication number
- CN108550047A CN108550047A CN201810231588.XA CN201810231588A CN108550047A CN 108550047 A CN108550047 A CN 108550047A CN 201810231588 A CN201810231588 A CN 201810231588A CN 108550047 A CN108550047 A CN 108550047A
- Authority
- CN
- China
- Prior art keywords
- data
- same day
- transactions
- total amount
- transaction
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0203—Market surveys; Market polls
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
This specification embodiment provides a kind of prediction technique and device of trading volume, in the prediction technique of trading volume, obtains the transaction data of same day last time and the relevant public sentiment data of transaction with the same day.Transaction data, public sentiment data and preset influence data are pre-processed, pretreated data are obtained.Pretreated data are inputted to the prediction model obtained according to historical data, with the total amount of transactions on the prediction same day.
Description
Technical field
This specification one or more embodiment is related to field of computer technology more particularly to a kind of prediction side of trading volume
Method and device.
Background technology
In order to provide preferably reply business demand, transaction platform usually requires before some day starts, to the same day
Total amount of transactions is predicted.In traditional technology, typically by time series models to the historical trading data before the same day into
Row study, to predict the total amount of transactions on the same day.
Accordingly, it is desirable to provide a kind of prediction technique of more accurate trading volume.
Invention content
This specification one or more embodiment describes a kind of prediction technique and device of trading volume, can improve prediction
Trading volume accuracy.
In a first aspect, a kind of prediction technique of trading volume is provided, including：
Obtain the first transaction data of same day last time and relevant first public sentiment data of transaction with the same day；
First transaction data, first public sentiment data and preset influence data are pre-processed, are obtained
Pretreated data；The preset data that influence refer to the data that the meeting pre-estimated has an impact trading volume；
By the pretreated data input prediction model, with the total amount of transactions on the prediction same day；The prediction model is
According to classical time series forecasting algorithm or machine learning algorithm, the second transaction data before the same day, the second public sentiment data and
What the preset influence data obtained；The prediction model is used to predict future development variation tendency according to historical data.
Second aspect provides a kind of prediction meanss of trading volume, including：
Acquiring unit, the first transaction data for obtaining same day last time and the transaction relevant first with the same day
Public sentiment data；
Pretreatment unit, first transaction data, first public sentiment data for being obtained to the acquiring unit
And preset influence data are pre-processed, and pretreated data are obtained；The preset influence data refer to estimating in advance
The data that the meeting of meter has an impact trading volume；
Predicting unit was used for the pretreated data input prediction model of the pretreatment unit, with the prediction same day
Total amount of transactions；The prediction model is according to classical time series forecasting algorithm or machine learning algorithm, the second friendship before the same day
What easy data, the second public sentiment data and the preset influence data obtained；The prediction model is used for according to historical data
To predict future development variation tendency.
The prediction technique and device for the trading volume that this specification one or more embodiment provides obtain same day last time
Transaction data and the relevant public sentiment data of transaction with the same day.On transaction data, public sentiment data and preset influence number
According to being pre-processed, pretreated data are obtained.Pretreated data are inputted to the prediction mould obtained according to historical data
Type, with the total amount of transactions on the prediction same day.It can thus be seen that when total amount of transactions of this specification on the day of prediction, consider
Work as day data and historical data, so as to improve prediction trading volume accuracy.
Description of the drawings
It is required in being described below to embodiment to make in order to illustrate more clearly of the technical solution of this specification embodiment
Attached drawing is briefly described, it should be apparent that, the accompanying drawings in the following description is only some embodiments of this specification, right
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings
Its attached drawing.
Fig. 1 is the schematic diagram for the forecasting system that this specification provides；
Fig. 2 is the prediction technique flow chart for the trading volume that this specification one embodiment provides；
Fig. 3 is the prediction meanss schematic diagram for the trading volume that this specification one embodiment provides.
Specific implementation mode
Below in conjunction with the accompanying drawings, the scheme provided this specification is described.
The prediction technique for the trading volume that this specification one embodiment provides can be applied to forecasting system as shown in Figure 1
In, in Fig. 1, forecasting system may include：Reptile module 101, real-time data synchronization module 102, synchronizing traffic data module
103, data preprocessing module 104 and prediction module 105.
Reptile module 101 for high-frequency crawls (e.g., every one minute) from webpage and relevant public sentiment number of merchandising
According to.The public sentiment data may include：The real-time live information (e.g., advertising campaign etc.) of trade company, fluctuation of exchange rate information and extreme
Weather information (e.g., extremely cold weather or very hot weather etc.) etc..
It should be noted that because when advertising campaign is done by trade company, it will usually the transaction count of user is greatly increased,
To which one day trading volume can be influenced, so when total amount of transactions of this specification on the day of prediction, the real-time activity of trade company is believed
Cease an effect characteristics as trading volume (also referred to as transaction amount).Further, since fluctuation of exchange rate situation and extreme weather are equal
One day trading volume can be influenced, so by the rwo also as the effect characteristics of trading volume.
By reptile module 101, trade company's action message can be known in advance.
Real-time data synchronization module 102 for high frequency obtained (e.g., every one minute) from start time on the same day (e.g., zero
Point) act the transaction data of every transaction occurred.Wherein, the transaction data of a transaction may include：Exchange hour, transaction
The amount of money, merchant information and user information etc..
Non- friendship that synchronizing traffic data module 103 is used to record in sync database, trading volume being had an impact
Easy data (hereinafter referred to as influencing data).Above-mentioned influence data may include：1) representation data of trade company：The country origin of trade company is believed
Breath, the type (e.g., entity trade company or virtual trade company etc.) of trade company, the business scope of trade company and the market in trade company location letter
It ceases (e.g., holiday information) etc..2) transaction stroke count.3) representation data (e.g., student, teacher and civilian worker etc.) etc. of user.
It should be noted that above-mentioned influence data can be collected artificially in advance.
Data preprocessing module 104 is used for above-mentioned public sentiment data (acquisition of reptile module 101), transaction data (number in real time
Obtained according to synchronization module 102) and influence data (acquisition of synchronizing traffic data module 103) progress data Layer fusion.Herein
Data Layer fusion can be understood as above-mentioned data being abstracted into higher-dimension from low-dimensional.Such as, by trade company's dimension, Yong Huwei in above-mentioned data
The data of degree and fund dimension are converted into trading volume latitude characterization.Data preprocessing module 104 is additionally operable to the number after merging
According to data format conversion is carried out, such as it is converted into and meets prediction model and enter to join the data format of standard, which may include
Data normalization etc..
Prediction module 105, for according to pretreated data, predicting the trading volume of some day.The prediction module 105 can
To include prediction model.The prediction model can be according to classical time series forecasting algorithm (or machine learning algorithm etc.) and history
What data obtained.Wherein, classical time series forecasting algorithm is a kind of for predicting the following generation using past and present data
The algorithm of variation tendency.It for example can be holt-winters algorithms (a kind of data smoothing algorithm), autoregression integral sliding
Average (Autoregressive Integrated Moving Average, ARIMA) algorithm and shot and long term memory network
(Long Short-Term Memory, LSTM) algorithm etc..And the transaction data before historical data may include the same day (is also referred to as
Historical trading data), the friendship public sentiment data (also referred to as history public sentiment data) before the same day and above-mentioned influence data etc..
It should be noted that above-mentioned prediction model can be an individual model, can also include multiple submodels.When
It, can be in conjunction with public sentiment data, to select corresponding submodel when including multiple submodels.Using public sentiment data as the real-time work of trade company
Dynamic information, prediction model include two submodels：For for first submodel and the second submodel.Assuming that the first submodel can
Being obtained according to the sample of classical time series forecasting algorithm (or machine learning algorithm etc.) and the real-time live information comprising trade company
；Second submodel can be according to classical time series forecasting algorithm (or machine learning algorithm etc.) and the reality not comprising trade company
When action message sample obtain.So when predicting the trading volume of some day, if the same day has trade company movable, it can pass through
First submodel is come the total amount of transactions on the day of predicting.If the same day does not have trade company movable, can be predicted by the second submodel
The total amount of transactions on the same day.Certainly, in practical applications, total transaction on the same day can also be predicted by two submodels simultaneously
Amount.Such as, respectively different weighted values is arranged in two submodels, later, according to the output valve of each submodel and corresponding
Weighted value, to predict the total amount of transactions on the same day.
It can thus be seen that when total amount of transactions of the above-mentioned forecasting system on the day of prediction, daylight trading number has been considered
According to, with the relevant public sentiment data of transaction on the same day and pre-estimate may to the data that trading volume has an impact, to
The accuracy of the trading volume of prediction can be improved.
Optionally, the forecasting system of Fig. 1 can also include merchant transaction real-time monitoring module 106, for high frequency to working as
Its transaction data is monitored, to determine whether corresponding trade company's activity.In one implementation, above-mentioned monitoring process can
Think：High frequency according to daylight trading data, count following data：Real trade amount, real trade stroke count and real trade
Objective unit price etc..Similarly, according to historical trading data, following data are counted：It history same period trading volume, historical trading stroke count and goes through
History visitor's unit price etc..It then can be respectively by real trade amount and history same period trading volume, real trade stroke count and historical trading pen
Number, real trade visitor's unit price are compared with history visitor's unit price.When it is both arbitrary differ greatly (e.g., be more than predetermined threshold value or
Less than predetermined threshold value) when, send out corresponding trade company's activity warning message.Such as, when more than predetermined threshold value, it is rapid to send out trade company's activity
Rise warning message.When less than predetermined threshold value, sends out trade company's activity and die-off warning message.
By merchant transaction real-time monitoring module 106, trade company's action message can be accurately identified.So that
Forecasting system considers trade company's action message when predicting trading volume, thus come further increase prediction trading volume standard
True property.
Optionally, the forecasting system of Fig. 1 can also include abnormality detection module 107.When further including abnormality detection module 107
When, trade company's activity warning message that merchant transaction real-time monitoring module 106 identifies can be input to abnormality detection module 107
In.
Abnormality detection module 107 can be used for carrying out abnormality detection the total amount of transactions on the same day, and for detect it is different
The total amount of transactions on the same day is adjusted in Chang Shi.In one implementation, above-mentioned abnormality detection and adjustment process can be with
For：According to historical trading data, following data are counted：Same period last week trading volume, this week are averaged trading volume and trading volume yesterday
Deng.Then the total amount of transactions on the same day and same period last week trading volume, this week be averaged trading volume and trading volume progress yesterday respectively
Compare one by one.If the total amount of transactions on the same day is higher than any of the above-described specified multiple (e.g., 1.5 times), and without the activity of corresponding trade company
Rise sharply warning message, it is determined that the total amount of transactions on the same day exists abnormal.Alternatively, if the total amount of transactions on the same day is less than any of the above-described finger
Determine multiple (e.g., 0.5 times), and die-off warning message without corresponding trade company's activity, it is determined that there are different for the total amount of transactions on the same day
Often.It is depositing when abnormal, adjustment factor can be multiplied by there will be the total amount of transactions on the abnormal same day (can be according to above-mentioned specified times
Number determines, e.g., 1.5 and 0.5), to ensure prediction the same day total amount of transactions in confidence band, can also ensure prediction
The same day total amount of transactions robustness.
Optionally, the forecasting system of Fig. 1 can also include external Policy Decision module 108, related to merchandising for collecting
Strategy.It, can with relevant strategy is merchandised for by taking the total amount of transactions on the same day is used to determine the scene of the demand of foreign exchange as an example
Think foreign exchange transaction strategy.The foreign exchange transaction strategy can be：The foreign exchange of this purchase only covers 60% position or whole friendship
Easily amount is less than true sale amount etc..
Optionally, the forecasting system of Fig. 1 can also include aid decision module 109.When further including aid decision module 109
When, what external Policy Decision module 108 was collected can be input to relevant strategy of merchandising in aid decision module 109.
Aid decision module 109 can be used for combining above-mentioned and merchandise relevant strategy, current transaction stream, history to hand over
The easy information such as trend and business KPI Key Performance Indicator (Key Performance Indicator, KPI), total friendship to the same day
Easily amount is adjusted.Such as previous example, it is assumed that relevant strategy is with merchandising：Whole trading volume is less than true sale amount, then may be used
The total amount of transactions on the same day is multiplied by penalty, to ensure to obtain optimal trading volume.
It should be noted that above-mentioned current transaction stream can be occurred according to from start time on the same day (e.g., 0 point)
Every transaction transaction data determined by.Above-mentioned historical trading trend and business KPI can be according to historical trading number
According to identified.For according to current transaction stream for the total amount of transactions on the same day is adjusted, if current transaction becomes
Gesture is ascendant trend, then the total amount of transactions on the same day can amplify according to preset ratio, otherwise similarly.Above-mentioned preset ratio can root
Ratio-dependent according to current trading volume (being determined according to the transaction data of same day last time) with current transaction amount.
Certainly, in practical applications, above-mentioned forecasting system can also include other modules, e.g., output module and displaying mould
Block, this specification are not construed as limiting this.
Fig. 2 is the prediction technique flow chart for the trading volume that this specification one embodiment provides.As shown in Fig. 2, the side
Method can specifically include：
Step 210, the first transaction data of same day last time and relevant first public sentiment of transaction with the same day are obtained
Data.
Herein, the first transaction data of same day last time can refer to produced by from start time on the same day to the current moment
Every transaction transaction data, be referred to as daylight trading data.Assuming that it is zero to be carved at the beginning of one day, and it is current
Moment is at 11 points in the morning, then the first transaction data of same day last time refers to the same day 00：00-11:It is generated every between 00
The transaction data of transaction.The transaction data may include：Exchange hour, transaction amount, merchant information and user information etc.
Deng.Specifically, can be that daylight trading data are obtained by real-time data synchronization module 102.
In addition, relevant first public sentiment data of the above-mentioned transaction with the same day (also referred to as same day public sentiment data) may include：Quotient
Family real-time live information, fluctuation of exchange rate information and extreme weather etc. can be obtained from webpage by reptile module 101
's.It should be noted that due to businessman would generally prior notice action message, so above-mentioned trade company's real-time live information can be
It is obtained according to the second public sentiment data (also referred to as history public sentiment data) before the same day.In short, this specification embodiment can carry
Before know trade company's action message.
Step 220, the first transaction data, the first public sentiment data and preset influence data are pre-processed, is obtained
Pretreated data.
Preset influence data herein can refer to it is above-mentioned be recorded in several storehouse tables, pre-estimate may to transaction
The nontransaction data having an impact are measured, i.e., are obtained by synchronizing traffic data module 103.Above-mentioned pretreatment may include data fusion
And data format conversion etc..
Specifically, can be by data prediction model 104 to the first transaction data, the first public sentiment data and preset
It influences data and carries out data Layer fusion.Data Layer fusion herein can be understood as above-mentioned data being abstracted into higher-dimension from low-dimensional.
Such as, the data of trade company's dimension, user's dimension and fund dimension in above-mentioned data are converted to trading volume latitude characterization.Later,
Data after fusion are subjected to data format conversion, is such as converted into and is met prediction model and enter to join the data format of standard, the conversion
Process may include data normalization etc..
Step 230, by pretreated data input prediction model, with the total amount of transactions on the prediction same day.
Prediction model herein may be embodied in prediction module 105, namely by total friendship on the prediction same day of prediction module 105
Yi Liang.Prediction model can be according to before the same day the second transaction data (historical trading data), history public sentiment data and
It is above-mentioned preset influencing data acquisition, it can be used for predicting future development variation tendency according to historical data.
It specifically, can be by the historical trading data collected in advance, history public sentiment data and above-mentioned preset influence number
Classical time series forecasting algorithm or machine learning algorithm etc. are trained according to as training sample.It is understood that will be classical
After time series forecasting algorithm or machine learning algorithm train, so that it may to obtain above-mentioned prediction model.It should be noted that instruction
The prediction model perfected can be an individual model, can also include multiple submodels.
When for an individual model, pretreated data are inputted into a model, so that it may to obtain the same day
Total amount of transactions.It, can be in conjunction with the first public sentiment data, to select corresponding submodel when including multiple submodels.It later will be pre-
The submodel for data input selection that treated obtains the total amount of transactions on the same day.Alternatively, pretreated data are inputted respectively
Into multiple submodels, multiple prediction trading volumes are obtained.Multiple prediction trading volumes are merged, total transaction on the same day is obtained
Amount.
Using the first public sentiment data as the real-time live information of trade company, prediction model includes two submodels：First submodel
For for the second submodel.Assuming that the first submodel can be according to classical time series forecasting algorithm (or machine learning calculate
Method etc.) and the sample of real-time live information comprising trade company obtain；Second submodel can be calculated according to classical time series forecasting
What the sample of method (or machine learning algorithm etc.) and the real-time live information not comprising trade company obtained.So predicting some day
Trading volume when, if the same day has, trade company is movable, and the total amount of transactions on the same day can be predicted by the first submodel.If the same day does not have
There is trade company's activity, then can predict the total amount of transactions on the same day by the second submodel.Certainly, in practical applications, can also
The total amount of transactions on the same day is predicted by two submodels simultaneously.Such as, respectively different weighted values is arranged in two submodels, it
Afterwards, according to the output valve of each submodel and corresponding weighted value, to predict the total amount of transactions on the same day.
It can thus be seen that when total amount of transactions of this specification embodiment on the day of prediction, daylight trading has been considered
Data, with the relevant public sentiment data of transaction on the same day and pre-estimate may to the data that trading volume has an impact, from
And the accuracy of the trading volume of prediction can be improved.
Optionally, during executing above-mentioned steps 210- steps 230, following process can also be performed simultaneously：Businessman
Transaction 106 high frequency of real-time monitoring module daylight trading data are monitored, to determine whether corresponding trade company's activity.
In a kind of realization method, above-mentioned monitoring process can be：High frequency according to daylight trading data, count following data：It is practical to hand over
Yi Liang, real trade stroke count and real trade visitor's unit price etc..Similarly, according to historical trading data, following data are counted：History
Same period trading volume, historical trading stroke count and history visitor's unit price etc..Then real trade amount and the history same period can be handed over respectively
Yi Liang, real trade stroke count are compared with historical trading stroke count, real trade visitor's unit price with history visitor's unit price.When both arbitrary
When differing greatly and (e.g., being more than predetermined threshold value or be less than predetermined threshold value), corresponding trade company's activity warning message is sent out.Such as, exist
When more than predetermined threshold value, sends out trade company's activity and rise sharply warning message.When less than predetermined threshold value, sends out trade company's activity and die-off alarm
Information.
By being monitored to daylight trading data, trade company's action message can be accurately identified.So as to pre-
Survey trading volume when, consider trade company's action message, thus come further increase prediction trading volume accuracy.
After the trading volume on the day of obtaining, the trade company that merchant transaction real-time monitoring module 106 can will identify that is movable
Warning message is input in abnormality detection module 107.To combine above-mentioned trade company's activity alarm signal by abnormality detection module 107
Breath, to be carried out abnormality detection to the total amount of transactions on the same day, and when an exception is detected, is adjusted the total amount of transactions on the same day.
In one implementation, above-mentioned abnormality detection and adjustment process can be：According to historical trading data, following number is counted
According to：Same period last week trading volume, this week are averaged trading volume and trading volume yesterday etc..Then respectively by the total amount of transactions on the same day with it is upper
All same period trading volumes, this week are averaged trading volume and trading volume yesterday is compared one by one.If the total amount of transactions on the same day is higher than upper
It states any specified multiple (e.g., 1.5 times), and rises sharply warning message without corresponding trade company's activity, it is determined that total transaction on the same day
Amount exists abnormal.Alternatively, if the total amount of transactions on the same day is less than any of the above-described specified multiple (e.g., 0.5 times), and no corresponding quotient
Family activity is die-offed warning message, it is determined that the total amount of transactions on the same day exists abnormal.It is depositing when abnormal, it can there will be abnormal
The total amount of transactions on the same day be multiplied by adjustment factor (can be determined according to above-mentioned specified multiple, e.g., 1.5 and 0.5), to ensure to predict
The same day total amount of transactions in confidence band.
Certainly, in practical applications, the total amount of transactions on the same day can also be carried out abnormality detection otherwise,
Such as, it can be carried out abnormality detection by case collimation method, this specification is not construed as limiting this.
In another implementation, it can also be collected by external Policy Decision module 108 and relevant strategy of merchandising.With
Can be that foreign exchange is handed over relevant strategy of merchandising for for the scene of demand of the total amount of transactions on the same day for determining foreign exchange
Easily strategy.The foreign exchange transaction strategy can be：The foreign exchange of this purchase only covers 60% position or whole trading volume less than true
Real trading volume etc..
Later, collection and relevant strategy of merchandising can be input to aid decision mould by external Policy Decision module 108
In block 109.Aid decision module 109 can in conjunction with the relevant strategy of merchandising, the total amount of transactions on the same day is adjusted.As before
State example, it is assumed that relevant strategy is with merchandising：Whole trading volume is less than true sale amount, then can be by the total amount of transactions on the same day
It is multiplied by penalty.To ensure to obtain optimal trading volume.
In another realization method, aid decision module 109 can be combined with preceding transaction stream, historical trading trend with
And the information such as business KPI Key Performance Indicator (Key Performance Indicator, KPI), the total amount of transactions on the same day is carried out
It adjusts.Current transaction stream can be according to determined by daylight trading data.Above-mentioned historical trading trend and business KPI can
To be according to determined by historical trading data.For the total amount of transactions on the same day to be adjusted according to current transaction stream come
It says, if current transaction stream is ascendant trend, the total amount of transactions on the same day can amplify according to preset ratio, otherwise similarly.
Above-mentioned preset ratio can be according to current trading volume (being determined according to the transaction data of same day last time) and current transaction amount
Ratio-dependent.
It should be noted that due to the outsourcing of line Shanghai, the fast development of business is paid under line face to face, Alipay supports trade company,
The Zhi Fuyu gatherings of different currency between buyer.The Alipay every workday all buys corresponding foreign exchange to cope with business demand,
Therefore, in the case where the accuracy of the trading volume of prediction improves, it is possible to reduce business risk and raising fund utilization rate.
To sum up, this specification above-described embodiment may be implemented in the case where there is trade company's moving obstacle, accurately to the same day
Total amount of transactions predicted.
Accordingly with the prediction technique of above-mentioned trading volume, a kind of trading volume that this specification one embodiment also provides is pre-
Device is surveyed, as shown in figure 3, the device includes：
Acquiring unit 301, for the first transaction data of last time on the day of obtaining and relevant with the transaction on the same day
First public sentiment data.
Pretreatment unit 302, the first transaction data, the first public sentiment data for obtaining to acquiring unit 301 and in advance
If influence data pre-processed, obtain pretreated data.The preset data that influence refer to the meeting pre-estimated to handing over
Easily measure the data having an impact.
Predicting unit 303 is used for 302 pretreated data input prediction model of pretreatment unit, to predict the same day
Total amount of transactions.Prediction model is according to classical time series forecasting algorithm or machine learning algorithm, the second transaction before the same day
What data, the second public sentiment data and preset influence data obtained.Prediction model is used to predict future according to historical data
Development tendency.
Predicting unit 303 specifically can be used for：
According to pretreated first public sentiment data, corresponding submodel is selected.
By the submodel of pretreated data input selection, the total amount of transactions on the same day is obtained.
Alternatively, pretreated data are separately input in multiple submodels, multiple prediction trading volumes are obtained.
Multiple prediction trading volumes are merged, the total amount of transactions on the same day is obtained.
It should be noted that above-mentioned acquiring unit 301 can be by the reptile module 101 and real-time data synchronization mould in Fig. 1
Block 102 is realized.Pretreatment unit 302 can realize by the data preprocessing module 104 in Fig. 1, and predicting unit 303 can be with
It is realized by the prediction module 105 in Fig. 1.
Optionally, which can also include：Comparing unit 304 and transmission unit 305.
Acquiring unit 301 is additionally operable to periodically according to the second transaction data, and the history same period for obtaining the first transaction data hands over
Easy data.
Comparing unit 304, the history same period transaction data for obtaining the first transaction data and acquiring unit 301 carry out
Compare.
Transmission unit 305, the difference for comparing the first transaction data and history same period transaction data when comparing unit 304
When larger, corresponding trade company's activity warning message is sent out.
It should be noted that above-mentioned comparing unit 304 and transmission unit 305 can in real time be supervised by the merchant transaction in Fig. 1
Module 106 is controlled to realize.
Optionally, which can also include：
Detection unit 306 is carried out abnormality detection for the total amount of transactions to the same day.
Unit 307 is adjusted, for detecting that the total amount of transactions on the same day is deposited when abnormal when detection unit 306, to the same day
Total amount of transactions is adjusted.
Optionally, detection unit 306 specifically can be used for：
According to the second transaction data, the history same period trading volume of the total amount of transactions on the same day is counted.
History same period trading volume is compared with the total amount of transactions on the same day.
When history same period trading volume is more than threshold value with the difference multiple of the total amount of transactions on the same day and lives without corresponding trade company
When dynamic warning message, it is abnormal to determine that the total amount of transactions on the same day exists.
Unit 307 is adjusted specifically to can be used for：
According to difference multiple, corresponding adjustment factor is determined.
According to adjustment factor, the total amount of transactions on the same day is adjusted.
Optionally, which can also include：
It should be noted that above-mentioned detection unit 306 and adjusting unit 307 can be realized by abnormality detection module 107.
Determination unit 308, for according to the first transaction data, determining current transaction stream.
Unit 307 is adjusted, the current transaction stream for being determined according to determination unit 308 adjusts the total amount of transactions on the same day.
And/or
Determination unit 308, for according to the second transaction data, determining historical trading trend.
Unit 307 is adjusted, the historical trading trend for being determined according to determination unit 308 adjusts the total amount of transactions on the same day.
And/or
Acquiring unit 301 is additionally operable to obtain external transaction strategy
Unit 307 is adjusted, the external transaction strategy for being obtained according to acquiring unit 301 adjusts the total amount of transactions on the same day.
It should be noted that above-mentioned determination unit 308 can be realized by the aid decision module 109 in Fig. 1.
The function of each function module of this specification above-described embodiment device can pass through each step of above method embodiment
Rapid to realize, therefore, the specific work process for the device that this specification one embodiment provides does not repeat again herein.
The prediction meanss for the trading volume that this specification one embodiment provides, acquiring unit 301 obtain same day last time
The first transaction data and relevant first public sentiment data of transaction with the same day.Pretreatment unit 302 to the first transaction data,
First public sentiment data and preset influence data are pre-processed, and pretreated data are obtained.Predicting unit 303 will be located in advance
Data input prediction model after reason, with the total amount of transactions on the prediction same day.Thus, it is possible to the accuracy of the trading volume of prediction.
Those skilled in the art are it will be appreciated that in said one or multiple examples, described in this specification
Function can be realized with hardware, software, firmware or their arbitrary combination.It when implemented in software, can be by these work(
Can storage in computer-readable medium or as on computer-readable medium one or more instructions or code passed
It is defeated.
Above-described specific implementation mode has carried out into one the purpose, technical solution and advantageous effect of this specification
Step is described in detail, it should be understood that the foregoing is merely the specific implementation mode of this specification, is not used to limit this
The protection domain of specification, all any modifications on the basis of the technical solution of this specification, made, change equivalent replacement
Into etc., it should all be included within the protection domain of this specification.
Claims (12)
1. a kind of prediction technique of trading volume, which is characterized in that including：
Obtain the first transaction data of same day last time and relevant first public sentiment data of transaction with the same day；
First transaction data, first public sentiment data and preset influence data are pre-processed, pre- place is obtained
Data after reason；The preset data that influence refer to the data that the meeting pre-estimated has an impact trading volume；
By the pretreated data input prediction model, with the total amount of transactions on the prediction same day；The prediction model is basis
Classical time series forecasting algorithm or machine learning algorithm, the second transaction data before the same day, the second public sentiment data and described
It is preset to influence what data obtained；The prediction model is used to predict future development variation tendency according to historical data.
2. according to the method described in claim 1, it is characterized in that, further including：
Periodically according to second transaction data, the history same period transaction data of first transaction data is obtained；
First transaction data is compared with history same period transaction data；
When first transaction data and when differing greatly of history same period transaction data, corresponding trade company's activity alarm signal is sent out
Breath.
3. according to the method described in claim 1, it is characterized in that, after the total amount of transactions on the day of obtaining, further include：
The total amount of transactions on the same day is carried out abnormality detection；
Total amount of transactions on the day of detecting is deposited when abnormal, and the total amount of transactions on the same day is adjusted.
4. according to the method described in claim 3, it is characterized in that, the total amount of transactions to the same day carries out abnormality detection, packet
It includes：
According to second transaction data, the history same period trading volume of the total amount of transactions on the same day is counted；
History same period trading volume is compared with the total amount of transactions on the same day；
When history same period trading volume is more than threshold value and no corresponding trade company's activity report with the difference multiple of the total amount of transactions on the same day
When alert information, it is abnormal to determine that the total amount of transactions on the same day exists；
The total amount of transactions to the same day is adjusted, including：
According to the difference multiple, corresponding adjustment factor is determined；
According to the adjustment factor, the total amount of transactions on the same day is adjusted.
5. according to the method described in claim 1, it is characterized in that, after the total amount of transactions on the day of described obtain, further include：
According to first transaction data, current transaction stream is determined；
According to the current transaction stream, the total amount of transactions on the same day is adjusted；
And/or according to second transaction data, determine historical trading trend；
According to the historical trading trend, the total amount of transactions on the same day is adjusted；
And/or obtain external transaction strategy；
According to the external transaction strategy, the total amount of transactions on the same day is adjusted.
6. according to claim 1-5 any one of them methods, which is characterized in that the prediction model includes multiple submodels,
The submodel is corresponding with first public sentiment data；
It is described by the pretreated data input prediction model, with the total amount of transactions on the prediction same day, including：
According to pretreated first public sentiment data, corresponding submodel is selected；
By the submodel of the pretreated data input selection, the total amount of transactions on the same day is obtained；
Alternatively, the pretreated data are separately input in the multiple submodel, multiple prediction trading volumes are obtained；
The multiple prediction trading volume is merged, the total amount of transactions on the same day is obtained.
7. a kind of prediction meanss of trading volume, which is characterized in that including：
Acquiring unit, the first transaction data for obtaining same day last time and relevant first public sentiment of transaction with the same day
Data；
Pretreatment unit, first transaction data, first public sentiment data for obtaining to the acquiring unit and
Preset influence data are pre-processed, and pretreated data are obtained；The preset influence data refer to pre-estimating
The data that trading volume can be had an impact；
Predicting unit is used for the pretreated data input prediction model of the pretreatment unit, with total friendship on the prediction same day
Yi Liang；The prediction model is according to classical time series forecasting algorithm or machine learning algorithm, the second number of deals before the same day
According to, the second public sentiment data and described preset influence what data obtained；The prediction model is used for according to historical data come pre-
Survey future development variation tendency.
8. device according to claim 7, which is characterized in that further include：Comparing unit and transmission unit；
The acquiring unit is additionally operable to periodically obtain the history of first transaction data according to second transaction data
Same period transaction data；
The comparing unit, the history same period number of deals for obtaining first transaction data and the acquiring unit
According to being compared；
Transmission unit, for the difference when the comparing unit first transaction data and history same period transaction data compared with
When big, corresponding trade company's activity warning message is sent out.
9. device according to claim 7, which is characterized in that further include：
Detection unit is carried out abnormality detection for the total amount of transactions to the same day；
Unit is adjusted, for detecting that the total amount of transactions on the same day is deposited when abnormal when the detection unit, total transaction to the same day
Amount is adjusted.
10. device according to claim 9, which is characterized in that
The detection unit is specifically used for：
According to second transaction data, the history same period trading volume of the total amount of transactions on the same day is counted；
History same period trading volume is compared with the total amount of transactions on the same day；
When history same period trading volume is more than threshold value and no corresponding trade company's activity report with the difference multiple of the total amount of transactions on the same day
When alert information, it is abnormal to determine that the total amount of transactions on the same day exists；
The adjusting unit is specifically used for：
According to the difference multiple, corresponding adjustment factor is determined；
According to the adjustment factor, the total amount of transactions on the same day is adjusted.
11. device according to claim 7, which is characterized in that further include：
Determination unit, for according to first transaction data, determining current transaction stream；
Unit is adjusted, the current transaction stream for being determined according to the determination unit adjusts the total amount of transactions on the same day；
And/or
Determination unit, for according to second transaction data, determining historical trading trend；
Unit is adjusted, the historical trading trend for being determined according to the determination unit adjusts the total amount of transactions on the same day；
And/or
The acquiring unit is additionally operable to obtain external transaction strategy；
Unit is adjusted, the external transaction strategy for being obtained according to the acquiring unit adjusts the total amount of transactions on the same day.
12. according to claim 7-11 any one of them devices, which is characterized in that the prediction model includes multiple submodules
Type, the submodel are corresponding with first public sentiment data；
The predicting unit is specifically used for：
According to pretreated first public sentiment data, corresponding submodel is selected；
By the submodel of the pretreated data input selection, the total amount of transactions on the same day is obtained；
Alternatively, the pretreated data are separately input in the multiple submodel, multiple prediction trading volumes are obtained；
The multiple prediction trading volume is merged, the total amount of transactions on the same day is obtained.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810231588.XA CN108550047A (en) | 2018-03-20 | 2018-03-20 | The prediction technique and device of trading volume |
| TW108100360A TWI690865B (en) | 2018-03-20 | 2019-01-04 | Transaction volume prediction method and device |
| PCT/CN2019/071874 WO2019179223A1 (en) | 2018-03-20 | 2019-01-16 | Transaction volume prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810231588.XA CN108550047A (en) | 2018-03-20 | 2018-03-20 | The prediction technique and device of trading volume |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108550047A true CN108550047A (en) | 2018-09-18 |
Family
ID=63516846
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810231588.XA Pending CN108550047A (en) | 2018-03-20 | 2018-03-20 | The prediction technique and device of trading volume |
Country Status (3)
| Country | Link |
|---|---|
| CN (1) | CN108550047A (en) |
| TW (1) | TWI690865B (en) |
| WO (1) | WO2019179223A1 (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109615449A (en) * | 2018-10-25 | 2019-04-12 | 阿里巴巴集团控股有限公司 | A kind of prediction technique and device, a kind of calculating equipment and storage medium |
| CN109672795A (en) * | 2018-11-14 | 2019-04-23 | 平安科技（深圳）有限公司 | Call center resource management method and device, electronic equipment, storage medium |
| CN110163752A (en) * | 2019-04-16 | 2019-08-23 | 阿里巴巴集团控股有限公司 | A kind of dealing amount of foreign exchange prediction technique, apparatus and system |
| CN110188793A (en) * | 2019-04-18 | 2019-08-30 | 阿里巴巴集团控股有限公司 | Data exception analysis method and device |
| WO2019179223A1 (en) * | 2018-03-20 | 2019-09-26 | 阿里巴巴集团控股有限公司 | Transaction volume prediction method and device |
| CN111598699A (en) * | 2020-05-20 | 2020-08-28 | 嘉联支付有限公司 | Divide moist system based on cluster |
| CN111709532A (en) * | 2020-05-26 | 2020-09-25 | 重庆大学 | Model-independent local interpretation-based online shopping representative sample selection system |
| WO2020211245A1 (en) * | 2019-04-19 | 2020-10-22 | 平安科技（深圳）有限公司 | Development trend data acquisition method and device |
| CN112651785A (en) * | 2020-12-31 | 2021-04-13 | 中国农业银行股份有限公司 | Real-time monitoring method and system for transaction amount |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI758676B (en) * | 2020-01-03 | 2022-03-21 | 華南商業銀行股份有限公司 | Financial transaction volume warning system |
| TWI765847B (en) * | 2020-01-03 | 2022-05-21 | 華南商業銀行股份有限公司 | Financial transaction volume warning system with promotion function |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8015454B1 (en) * | 2008-06-02 | 2011-09-06 | Quest Software, Inc. | Computer systems and methods for predictive performance management of data transactions |
| CN103577413B (en) * | 2012-07-20 | 2017-11-17 | 阿里巴巴集团控股有限公司 | Search result ordering method and system, search results ranking optimization method and system |
| CN104424598A (en) * | 2013-09-06 | 2015-03-18 | 株式会社日立制作所 | Cash demand quantity predicating device and method |
| CN104408558B (en) * | 2014-11-18 | 2017-08-25 | 南方电网科学研究院有限责任公司 | A kind of accounting operations analogue simulation model system |
| CN104778622A (en) * | 2015-04-29 | 2015-07-15 | 清华大学 | Method and system for predicting TPS transaction event threshold value |
| TWI578242B (en) * | 2015-11-13 | 2017-04-11 | 財團法人資訊工業策進會 | Technology trend predicting method and system and non-transitory computer readable storage medium |
| CN107067283B (en) * | 2017-04-21 | 2021-05-18 | 重庆邮电大学 | E-commerce consumption customer flow prediction method based on historical merchant records and user behaviors |
| 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 |
-
2018
- 2018-03-20 CN CN201810231588.XA patent/CN108550047A/en active Pending
-
2019
- 2019-01-04 TW TW108100360A patent/TWI690865B/en active
- 2019-01-16 WO PCT/CN2019/071874 patent/WO2019179223A1/en active Application Filing
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019179223A1 (en) * | 2018-03-20 | 2019-09-26 | 阿里巴巴集团控股有限公司 | Transaction volume prediction method and device |
| CN109615449A (en) * | 2018-10-25 | 2019-04-12 | 阿里巴巴集团控股有限公司 | A kind of prediction technique and device, a kind of calculating equipment and storage medium |
| CN109672795B (en) * | 2018-11-14 | 2022-03-11 | 平安科技（深圳）有限公司 | Call center resource management method and device, electronic equipment and storage medium |
| CN109672795A (en) * | 2018-11-14 | 2019-04-23 | 平安科技（深圳）有限公司 | Call center resource management method and device, electronic equipment, storage medium |
| CN110163752A (en) * | 2019-04-16 | 2019-08-23 | 阿里巴巴集团控股有限公司 | A kind of dealing amount of foreign exchange prediction technique, apparatus and system |
| CN110188793A (en) * | 2019-04-18 | 2019-08-30 | 阿里巴巴集团控股有限公司 | Data exception analysis method and device |
| CN110188793B (en) * | 2019-04-18 | 2024-02-09 | 创新先进技术有限公司 | Data anomaly analysis method and device |
| WO2020211245A1 (en) * | 2019-04-19 | 2020-10-22 | 平安科技（深圳）有限公司 | Development trend data acquisition method and device |
| CN111598699A (en) * | 2020-05-20 | 2020-08-28 | 嘉联支付有限公司 | Divide moist system based on cluster |
| 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 |
| CN112651785A (en) * | 2020-12-31 | 2021-04-13 | 中国农业银行股份有限公司 | Real-time monitoring method and system for transaction amount |
| CN112651785B (en) * | 2020-12-31 | 2023-12-08 | 中国农业银行股份有限公司 | Transaction amount real-time monitoring method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| TWI690865B (en) | 2020-04-11 |
| TW201941120A (en) | 2019-10-16 |
| WO2019179223A1 (en) | 2019-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108550047A (en) | The prediction technique and device of trading volume | |
| CN108764863B (en) | Data transfer method, device, server and storage medium | |
| WO2019001120A1 (en) | Method and system for processing dynamic pricing data of commodity | |
| CN108228706A (en) | For identifying the method and apparatus of abnormal transaction corporations | |
| KR102009309B1 (en) | Management automation system for financial products and management automation method using the same | |
| US9940673B2 (en) | Systems and methods for measuring relationships between investments and other variables | |
| CN109685651A (en) | The credit risk monitoring method and device of current assets mortgage | |
| CN111008896A (en) | Financial risk early warning method and device, electronic equipment and storage medium | |
| CN109166034A (en) | A kind of Risk Forecast Method and system | |
| CN109934268A (en) | Abnormal transaction detection method and system | |
| CN109583729B (en) | Data processing method and device for platform online model | |
| JPWO2017221448A1 (en) | INFORMATION PROCESSING APPARATUS, INFORMATION PROCESSING METHOD, AND PROGRAM | |
| CN103310353A (en) | Anti-attack data filtering optimization system and method | |
| CN109934700A (en) | A kind of method and device of arbitrage detecting | |
| CN109658643B (en) | Payment risk warning method, server and intelligent POS (point of sale) | |
| CN106557842A (en) | The method and device of prediction K lines | |
| CN107480703A (en) | Transaction fault detection method and device | |
| Baker | Risk aversion in maintenance: overmaintenance and the principal-agent problem | |
| CN115439265A (en) | Intelligent insurance industry compensation abnormal transaction risk control system | |
| CN108921607A (en) | A kind of Retail commodity Dynamic Pricing method, apparatus, equipment, system and medium | |
| CN112037071B (en) | Risk monitoring method and device for investment combination product | |
| TWI720638B (en) | Deposit interest rate bargaining adjustment system and method thereof | |
| KR102153834B1 (en) | Method and quantifying a data based on final value and estimate | |
| CN108470227A (en) | Based on the Regional Macro economic System prediction modeling method of assessment GDP growth figures and application | |
| CN113763032B (en) | Commodity purchase intention recognition method and device |
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: 20180918 |