CN109993339A - A kind of prediction technique for the financial business potential user that goes abroad - Google Patents
A kind of prediction technique for the financial business potential user that goes abroad Download PDFInfo
- Publication number
- CN109993339A CN109993339A CN201711483255.8A CN201711483255A CN109993339A CN 109993339 A CN109993339 A CN 109993339A CN 201711483255 A CN201711483255 A CN 201711483255A CN 109993339 A CN109993339 A CN 109993339A
- Authority
- CN
- China
- Prior art keywords
- abroad
- financial business
- user
- time series
- characteristic
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000000714 time series forecasting Methods 0.000 claims abstract description 50
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 40
- 238000012300 Sequence Analysis Methods 0.000 claims abstract description 13
- 238000012545 processing Methods 0.000 claims description 19
- 238000004458 analytical method Methods 0.000 claims description 16
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical group C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 claims description 16
- 238000003909 pattern recognition Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 3
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims description 2
- 238000003066 decision tree Methods 0.000 claims description 2
- 238000007637 random forest analysis Methods 0.000 description 4
- 238000005065 mining Methods 0.000 description 3
- 230000001737 promoting effect Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 230000007306 turnover Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Human Resources & Organizations (AREA)
- Technology Law (AREA)
- Data Mining & Analysis (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of prediction technique (200) for the financial business potential user that goes abroad, the personal characteristics data of financial business potential user this method comprises: the candidate of acquisition (201) within the time cycle occurred goes abroad, wherein the personal characteristics includes time series forecasting feature, invariant features and it is contemplated that feature;Characteristic trend curve model of these time series forecasting features in future time period is obtained by time sequence analysis algorithm by the go abroad time series forecasting characteristic of financial business potential user of the candidate of the acquisition, wherein obtaining corresponding characteristic data value (202) of the time series forecasting feature in future time period by the characteristic trend curve model;Using the candidate go abroad the prediction of financial business potential user time series forecasting characteristic value is in conjunction with corresponding invariant features value and it is contemplated that characteristic value by means of predict that the candidate goes abroad financial business potential user using disaggregated model based on the financial business of going abroad in corresponding future time period that financial business of going abroad promotes that user establishes is that will use or will be without using going abroad financial business user (203).
Description
Technical field
The present invention relates to finance data analysis fields, more particularly to the prediction side of financial business potential user that goes abroad a kind of
Method and related device.
Background technique
Document CN107093101A discloses a kind of potential loan usage mining based on POS pipelined data and risk score side
Method.Wherein embodiment of the document based on POS pipelined data to trade company for fund and demand for loan amount, is flowed by using POS
The user characteristic data of water differentiates whether character pair meets condition to obtaining potential loan user using threshold rule, from
And it combines POS pipelined data in terms of expanding operation and starts in terms of capital turnover and realize potential loan usage mining and carry out
Risk assessment.
Document CN105761112A discloses the method for digging that pipe target customer was raised stocks and provided in a kind of securities finance.Wherein this article
It offers based on fund-raising gap business, is extracted from database and analyze user's association attributes, including fund-raising gap frequency, amount of money range
Equal ASSOCIATE STATISTICSs feature establishes Random Forest model and carries out target customer's identification using random forests algorithm, for judging
Potential user's object in non-targeted client.
Document CN106846061A discloses a kind of potential user's method for digging and device.Wherein the house property based on user is searched
Rope behavioural information is trained using the keyword feature extracting method analyzed based on URL and using machine learning algorithm according to preparatory
Obtained potential user's mining model predicts corresponding characteristic information, to carry out the judgement of potential the commercial house user.
The potential user for the time series dimension based on user characteristics that there are no relevant in currently available technology predicts hair
Existing method.Current business scene objects user's screening is that Classification and Identification simply based on user's Figure Characteristics or rule are sentenced
Not, characteristic dimension only includes a few features such as time, money, does not consider the variation feelings of the attributive character dimension at any time of user
Condition can not carry out the discovery of potential increment client and the formulation of corresponding marketing strategy.
Summary of the invention
Therefore, according to the first aspect of the invention, the prediction technique of financial business potential user that goes abroad a kind of, the party are proposed
Method includes:
Candidate within the time cycle occurred is obtained to go abroad the personal characteristics data of financial business potential user, wherein
The personal characteristics includes time series forecasting feature, invariant features and it is contemplated that feature;
By the candidate of the acquisition go abroad financial business potential user time series forecasting characteristic by time sequence
Column parser, such as ARIMA algorithm obtain characteristic trend curve of these time series forecasting features in future time period
Model, wherein it is corresponding in future time period to obtain the time series forecasting feature by the characteristic trend curve model
Characteristic data value;
Using the candidate go abroad the prediction of financial business potential user time series forecasting characteristic value combine it is corresponding
Invariant features value and it is contemplated that characteristic value by means of based on go abroad financial business promote user establish in corresponding future time
In period go abroad financial business using disaggregated model predict that the candidate goes abroad financial business potential user be will use or
It will be without using the financial business user that goes abroad.
The design of first aspect proposed by the present invention is, the candidate financial business potential user that goes abroad is occurring first
Time cycle in time series forecasting characteristic predicted by means of time sequence analysis algorithm, such as ARIMA algorithm
To obtain the corresponding time series forecasting characteristic data value in future time period, then utilize in future time period
The corresponding time series forecasting characteristic data value of prediction combines other corresponding characteristic uses in corresponding future time period
Interior financial business of going abroad is that will use or will be without using going out using the disaggregated model predicting candidate financial business potential user that goes abroad
State financial business user, so that the potential crowd for promoting client for financial business of going abroad can potentially be developed by searching out.
According to the present invention, the time cycle can be understood as selectable.The selection of time cycle on the one hand can with point
The analysis period is related, on the other hand can be related with business demand.Such as can monthly predict monthly situation, to be well-suited for
The marketing of some months is developed programs below;It can similarly predict to provide reference for the marketing volume in subsequent time per year.Therefore, according to
The present invention, time cycle may include the time spans such as the moon, season, half a year, year, 2 years.
According to the present invention, the candidate financial business potential user that goes abroad be can be understood as predicting that financial business of going abroad is potential
The candidate user collection of user, and the candidate user collection is typically from the user not being pushed using finance of going abroad.Have
Sharp ground, sets, which can be most of characteristic sequence curves in its time series forecasting feature and accord with herein
Close the user of sequence stationary.Preferably, it can be interpreted as 70% features above by most of herein and meet sequence stationary.This
If it is to be understood that not being pushed, using the financial user that goes abroad, there are four time series forecasting features altogether, then wherein extremely
To meet sequence stationary there are three feature less just and can be listed in candidate user is concentrated and to become candidate's financial business of going abroad potential
User.
Meeting stationarity according to sequence of the present invention is the premise for doing time series analysis.For example it is assumed that sometime
Sequence is generated by a certain random process, such as according to time series forecasting feature of the present invention.If the time series
Meeting its mean value is the constant unrelated with time t, and variance is the constant being unrelated with the time, and covariance is only related with time interval
The constant being unrelated with the time, then such Random time sequence is stable.
According to the present invention, if acquisition can be detected by Dicky-Fuller by meeting sequence stationary.Dicky-
Differentiation p value can be calculated in Fuller detection, indicate to meet sequence stationary if p value is less than 0.01.Time series is flat
Stability is the premise analyzed using time sequence analysis algorithm.Naturally, other it is any can be with cycle tests stationarity
Method is all admissible.
According to the present invention, so-called time series forecasting feature can be understood as be exactly this feature of user analog value not
The value come in the period is predictable according to the corresponding value of this feature in the time cycle occurred but cannot have no
It uniquely determines to problem, such time series forecasting feature may include for example: transaction of the user in specific period is total
The amount of money, the transaction frequency, single average deal size, ending balance etc..So-called invariant features can be understood as the relatively stable base of user
Constant feature in sheet, such as education degree (such as may include undergraduate course or less, undergraduate course, postgraduate or more).It is so-called can be pre-
Meter feature can be understood as the feature that can be relatively easily expected, such as home background and age range according to the present invention.It presses
According to the present invention, home background may include unmarried and non-unmarried;Age range can be divided into such as including (20 years old or less, 20-30
Year, 30-40 years old, 40-45 years old, 45-50 years old, 50-60 years old, 60 years old or more).Certainly if it is possible, other division modes are also
It is admissible.
Time sequence analysis algorithm is the mode analyzed one group of data.According to the present invention, week time herein
As soon as the phase is for only one value of time series forecasting feature or numerical value, therefore time sequence analysis algorithm needs multiple weeks
The data of phase form one group of value, in other words a vector being made of the value or numerical value of multiple time series forecasting features,
Operation could be carried out by respective algorithms.The multiple not stringent limitation in fact, generally higher than 50.
According to the present invention, ARIMA algorithm can be used as a kind of typical time sequence analysis algorithm.ARIMA algorithm
Also known as ARIMA model algorithm, full name are that autoregression integrates moving average model (Autoregressive Integrated
Moving Average Model is abbreviated ARIMA).So-called ARIMA model, refers to and converts nonstationary time series to steadily
Time series, then by dependent variable, only the present worth to its lagged value and stochastic error and lagged value are returned and are established
Model.
According to the present invention, user is promoted in financial business of going abroad can be understood as the use that financial business of going abroad is promoted to it
Family usually may include using the user of the user for financial business of going abroad and unused financial business of going abroad.
According to the present invention, it is described based on go abroad financial business promote user establish in corresponding future time period
Financial business of going abroad using disaggregated model can be used for that candidate going abroad financial business potential user in future time period
It is divided into and is namely predicted as that the user for financial business of going abroad will be used and the user for financial business of going abroad will not used.
Advantageously, described that the going abroad in corresponding future time period that user establishes is promoted based on financial business of going abroad
Financial business may include following sub-step using the foundation of disaggregated model:
Personal characteristics data of the user within the time cycle occurred are promoted in financial business of going abroad described in acquisition, wherein institute
Stating personal characteristics data may include time series forecasting feature, invariant features and it is contemplated that feature;
Based on the financial business of going abroad promote personal characteristics data of the user within the time cycle occurred by means of
The financial business of going abroad that pattern recognition classifier algorithm was established within the time cycle occurred uses disaggregated model, institute as above
It states, it may include using the user for financial business of going abroad and unused financial circles of going abroad that user is promoted in the financial business of going abroad
The user of business, that is to say, that by it is described go abroad financial business using disaggregated model can will go abroad financial business promote user
It is divided into and has used user and unused user.Advantageously, it in the model foundation, determines respectively for not making using user and
With the threshold value of the corresponding personal characteristics data of user;
The financial business of going abroad within the time cycle occurred based on foundation is using disaggregated model by means of time sequence
Column parser, such as ARIMA algorithm obtain the financial business of going abroad accordingly in future time period and use disaggregated model.Tool
Body as according to identical principle above, use disaggregated model to by the financial business of going abroad within the time cycle occurred
The time series forecasting feature in the corresponding personal characteristics data for having used user and unused user determined respectively
Threshold value obtains the corresponding threshold value in future time period using time sequence analysis algorithm, such as ARIMA algorithm, and is tying
It closes invariant features and it is contemplated that on the basis of feature, the financial business of going abroad established in future time period uses classification mould
Type, to this it can be appreciated that financial business of going abroad in future time period using disaggregated model is to utilize time series point
Analysis algorithm uses the financial business of going abroad within the time cycle occurred a kind of reconstruct of disaggregated model.Naturally, may be used
The financial business of going abroad in future time period is interpreted as going out within the time cycle occurred using disaggregated model
State's financial business uses disaggregated model following model.
Advantageously, the financial business of going abroad within the time cycle occurred uses in the foundation of disaggregated model, can
To determine the threshold value for having used the corresponding personal characteristics data of user and unused user respectively.
It can be advantageous to which thus obtaining the financial business of going abroad accordingly in future time period uses classification mould
Type, i.e., based on the determination for having used the personal characteristics data threshold of user and unused user by means of time series
Parser obtains the corresponding personal characteristics data threshold in future time period.
Advantageously, pattern recognition classifier algorithm typically may comprise optional random forest, decision tree etc. and other
Any machine learning algorithm that can determine classification thresholds.
Be according to the advantages of method proposed by the present invention: the present invention is based on time sequence analysis algorithms, not to user
Come the period time series forecasting feature predicted accordingly, so as to effectively avoid because year time change changing features
Caused by predict error.To analyze the possibility that user meets financial service of going abroad using the discrimination model in future period
Property, and targetedly service plan is provided according to the period forecasting of most possible property.
In addition, second aspect according to the invention also proposes the prediction dress of financial business potential user that goes abroad accordingly a kind of
It sets, which includes:
Acquiring unit consists of, potential for obtaining the financial business of going abroad of the candidate within the time cycle occurred
The personal characteristics data of user, wherein the personal characteristics includes time series forecasting feature, invariant features and it is contemplated that feature;
Analysis and processing unit consists of, and is gone abroad the potential use of financial business by the candidate obtained by the acquiring unit
The time series forecasting characteristic at family obtains these time series forecastings by time sequence analysis algorithm, such as ARIMA algorithm
Characteristic trend curve model of the feature in future time period, wherein obtaining the time by the characteristic trend curve model
Corresponding characteristic data value of the sequence prediction feature in future time period;
In addition, analysis and processing unit may be also constructed to, gone abroad the prediction of financial business potential user using the candidate
Time series forecasting characteristic value combine corresponding invariant features value and it is contemplated that characteristic value by means of based on going abroad financial business
It promotes the financial business of going abroad in corresponding future time period that user establishes and predicts that the candidate goes out using disaggregated model
State financial business potential user's is that will use or will not use the financial business user that goes abroad;
Wherein the acquiring unit can be communicated to connect via wired or wireless way and analysis and processing unit.
According to the present invention, acquiring unit can be understood as any type of data input and acquisition device, available
It numerical data and/or analogue data and is deposited by other wired or wireless communication modes with other data processing units or data
Storage unit is communicated, for use in the processing and storage of data.
According to the present invention, analysis and processing unit can be understood as any type of central processing unit or processing unit,
It can receive data and/or signal and it handled by corresponding algorithm or software, it furthermore can also be with output phase
The control signal answered is to controlled device or displays signal to corresponding display.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described.It should be evident that the accompanying drawings in the following description only describes
A part of the embodiments of the present invention.These attached drawings are not restrictive for the present invention, but are served illustrative.
Wherein:
Fig. 1 is shown according to a kind of flow chart of the prediction technique 200 for the financial business potential user that goes abroad proposed by the present invention;
Fig. 2 shows according to a kind of signal side of the prediction meanss 100 for the financial business potential user that goes abroad proposed by the present invention
Block diagram.
Specific embodiment
Fig. 1 is shown according to a kind of flow chart of the prediction technique 200 for the financial business potential user that goes abroad proposed by the present invention.
This method 200 includes the following steps:
Candidate within the time cycle occurred is obtained to go abroad the personal characteristics data 201 of financial business potential user,
Wherein the personal characteristics includes time series forecasting feature, invariant features and it is contemplated that feature;
By the candidate of the acquisition go abroad financial business potential user time series forecasting characteristic by time sequence
Column parser, such as ARIMA algorithm obtain characteristic trend curve of these time series forecasting features in future time period
Model, wherein it is corresponding in future time period to obtain the time series forecasting feature by the characteristic trend curve model
Characteristic data value 202;
Using the candidate go abroad the prediction of financial business potential user time series forecasting characteristic value combine it is corresponding
Invariant features value and it is contemplated that characteristic value by means of based on go abroad financial business promote user establish in corresponding future time
In period go abroad financial business using disaggregated model predict that the candidate goes abroad financial business potential user be will use or
It will be without using the financial business user 203 that goes abroad.
Below according to a specific embodiment to the prediction side according to the financial business potential user proposed by the present invention that goes abroad
Method 200 is elaborated.
Firstly, for the financial business user that goes abroad is not promoted, respectively by the period of 2010-2016 trade total amount,
The time series forecastings characteristic use Dicky-Fuller detection methods such as the frequency, single average deal size, ending balance of trading inspection
Whether the characteristic time sequence for surveying these users meets sequence stationary, if 70% or more features described above meets (i.e. four
At least meet sequence stationary there are three feature in feature) then retain user and goes abroad financial business potential user as candidate, such as
Fruit is unsatisfactory for, and rejects.
Then, candidate is selected in financial business user go abroad after financial business potential user never promoting to go abroad, phase
Also obtain with answering these user 2010-2016 invariant features such as education degree and as it is contemplated that feature family status
And age range.
Then, its total amount of trading, the transaction frequency, average transaction amount and friendship are utilized for candidate user elected
Ending balance is pre- as the time series that time series forecasting feature construction ARIMA model analysis obtains 2017,2018,2019
Feature is surveyed, while the corresponding invariant features value (education degree) determined in user in 2017,2018,2019 and it is contemplated that spy
Value indicative (home background, age range).Wherein, invariant features value is to be considered as constant, such as education journey after being registered as bank-user
It is postgraduate that degree is that postgraduate is then considered as always;It is contemplated that in feature home background characteristic value be user at 30 years old or more the value is
Non- unmarried, the value is consistent with home background when user's registration within 30 years old or less, if unmarried, is still if non-unmarried to be unmarried
It is non-unmarried.Age range is divided by actual age.According to the present embodiment, illustrative certain user 2017,2018,2019 pre-
Surveying feature can be as shown in table 1 below:
Table 1
Then, by means of promoting the financial circles of going abroad in 2017-2019 that user establishes based on financial business of going abroad
Make suring, predict that above-mentioned candidate goes abroad financial business potential user with disaggregated model is that will use or will be without using finance of going abroad
Service-user.
The prediction result of certain example user's property is shown as according to the present embodiment, such as the following table 2.
Table 2
Wherein, according to the present embodiment, financial business of going abroad in 2017-2019 includes such as using the acquisition of disaggregated model
Lower sub-step:
Firstly, obtaining financial business of going abroad promotes personal characteristics data of the user in the time occurred, it is same these
Personal characteristics data include time series forecasting feature, invariant features and it is contemplated that feature.Wherein, financial business of going abroad, which is promoted, to be used
Family includes using the user of the user for financial business of going abroad and unused financial business of going abroad.It goes abroad as the following table 3 shows part
Personal characteristics data cases of the user in 2010 are promoted in financial business:
Table 3
Then, based on it is above-mentioned go abroad financial business promote user occurred year in one's duty personal characteristics data by means of
Random forest discrimination model as pattern recognition classifier algorithm is established in the in one's duty financial business of going abroad of the above-mentioned year occurred
Using disaggregated model, it is referred to as classifier.
Specifically, in the year occurred in one's duty foundation of the financial business using disaggregated model of going abroad, determining pair respectively
In the threshold value for the corresponding personal characteristics data for having used user and unused user.
For example, the financial business discrimination model threshold of going abroad in table 4 shows individual features during 2010-2016 each year
Value.
Time | Transaction total amount in period | The transaction frequency | Single average deal size | Ending balance |
2010 | 25464 | 343 | 74.2 | 35683 |
2011 | 25876 | 354 | 73.1 | 54723 |
2012 | 25684 | 367 | 70.0 | 73289 |
2013 | 25793 | 235 | 109.8 | 64389 |
2014 | 26731 | 343 | 77.9 | 83211 |
2015 | 26432 | 382 | 69.2 | 109281 |
2016 | 26639 | 412 | 64.7 | 153453 |
Table 4
Then, based on it is determining it is above-mentioned for used the personal characteristics data threshold of user and unused user by means of
It is corresponding personal characteristics data threshold in 2017-2019 that ARIMA algorithm, which obtains in future time period,.
It is as follows that table 5 schematically illustrates all kinds of characteristic threshold values for predicting to obtain 2017,2018,2019 by ARIMA algorithm:
Time | Transaction total amount in period | The transaction frequency | Single average deal size | Ending balance |
2017 | 27311 | 397 | 73.6 | 182342.4 |
2018 | 27541 | 412 | 79.1 | 201123.6 |
2019 | 27843 | 416 | 76.4 | 213489.2 |
Table 5
Finally, these threshold values are imparted in classifier again, building corresponding 2017,2018,2019 is circannian to go abroad
Financial business uses user's identification model.
Fig. 2 shows according to a kind of signal side of the prediction meanss 100 for the financial business potential user that goes abroad proposed by the present invention
Block diagram.The device 100 includes:
Acquiring unit 101, consists of, latent for obtaining the financial business of going abroad of the candidate within the time cycle occurred
In the personal characteristics data of user, wherein the personal characteristics includes time series forecasting feature, invariant features and it is contemplated that spy
Sign;
Analysis and processing unit 102, consists of, potential by the candidate obtained by the acquiring unit financial business of going abroad
It is pre- that the time series forecasting characteristic of user by time sequence analysis algorithm, such as ARIMA algorithm obtains these time serieses
Characteristic trend curve model of the feature in future time period is surveyed, wherein when obtaining described by the characteristic trend curve model
Between corresponding characteristic data value of the sequence prediction feature in future time period;
In addition, analysis and processing unit 102 may be also constructed to, gone abroad the pre- of financial business potential user using the candidate
The time series forecasting characteristic value of survey combine corresponding invariant features value and it is contemplated that characteristic value by means of based on going abroad financial circles
Business promotes the financial business of going abroad in corresponding future time period that user establishes and predicts the candidate using disaggregated model
Go abroad financial business potential user is that will use or will not use the financial business user that goes abroad;
Wherein the acquiring unit 101 can be communicated to connect via wired or wireless way and analysis and processing unit 102.
According to the present invention, acquiring unit 101 can be understood as any type of data input and acquisition device, can obtain
Access digital data and/or analogue data and pass through other wired or wireless communication modes and other data processing units or data
Storage unit is communicated, for use in the processing and storage of data.
According to the present invention, analysis and processing unit 102 can be understood as any type of central processing unit or processing unit,
It can receive data and/or signal and is handled by corresponding algorithm or software it, furthermore can also export
Corresponding control signal to controlled device or displays signal to corresponding display.
Above description to the embodiment proposed, enables those skilled in the art to implement or use the present invention.
It should be appreciated that the feature disclosed in above embodiments individually or can be tied mutually other than the situation for having special instruction
Ground is closed to use.Various modifications to these embodiments will be readily apparent to those skilled in the art, herein
Defined in General Principle can realize in other embodiments without departing from the spirit or scope of the present invention.
Therefore, invention disclosed herein is not limited to disclosed specific embodiment, but is intended to appended right such as and wants
Ask the modification within the spirit and scope of the present invention defined by book.
Claims (14)
1. a kind of prediction technique (200) for the financial business potential user that goes abroad, this method comprises:
(201) candidate within the time cycle occurred is obtained to go abroad the personal characteristics data of financial business potential user,
Described in personal characteristics include time series forecasting feature, invariant features and it is contemplated that feature;
Time series point is passed through by the go abroad time series forecasting characteristic of financial business potential user of the candidate of the acquisition
Analysis algorithm obtains characteristic trend curve model of these time series forecasting features in future time period, wherein by the spy
Sign trend curve model obtains corresponding characteristic data value of the time series forecasting feature in future time period
(202);
Using the candidate go abroad the prediction of financial business potential user time series forecasting characteristic value combine it is corresponding constant
Characteristic value and it is contemplated that characteristic value by means of based on go abroad financial business promote user establish in corresponding future time period
What interior financial business of going abroad predicted that the candidate goes abroad financial business potential user using disaggregated model is that will use or will not
Use go abroad financial business user (203).
2. prediction technique according to claim 1, which is characterized in that the time sequence analysis algorithm is ARIMA algorithm.
3. prediction technique according to claim 1, which is characterized in that the time cycle may be selected to be the moon, season, half
Year, Nian Huo 2 years.
4. prediction technique according to claim 1, which is characterized in that the candidate financial business potential user that goes abroad comes from
It in the user not being pushed using finance of going abroad and is that 70% features above sequence is bent in its all time series forecasting feature
Line meets the user of sequence stationary.
5. prediction technique according to claim 4, which is characterized in that whether meet sequence stationary and pass through Dicky-
Fuller detection obtains.
6. prediction technique according to claim 1, which is characterized in that the time series forecasting feature includes user in spy
Transaction total amount, the transaction frequency, single average deal size, ending balance in fixed cycle.
7. prediction technique according to claim 1, which is characterized in that the invariant features are education degrees.
8. prediction technique according to claim 1, which is characterized in that described it is contemplated that feature includes home background and age
Section.
9. prediction technique according to claim 1, which is characterized in that it includes having made that user is promoted in the financial business of going abroad
With the user for financial business of going abroad and the user of unused financial business of going abroad.
10. prediction technique according to claim 1, which is characterized in that described to be built based on the financial business distribution user that goes abroad
The vertical financial business of going abroad in corresponding future time period includes following sub-step using the foundation of disaggregated model:
Personal characteristics data of the user within the time cycle occurred are promoted in financial business of going abroad described in acquisition, wherein described
People's characteristic includes time series forecasting feature, invariant features and it is contemplated that feature;
Personal characteristics data of the user within the time cycle occurred are promoted by means of mode based on the financial business of going abroad
The financial business of going abroad that identification sorting algorithm was established within the time cycle occurred uses disaggregated model;
The financial business of going abroad within the time cycle occurred based on foundation is using disaggregated model by means of time series point
Analysis algorithm obtains the financial business of going abroad accordingly in future time period and uses disaggregated model.
11. prediction technique according to claim 10, which is characterized in that going abroad within the time cycle occurred
Financial business determines the corresponding personal characteristics for having used user and unused user using in the foundation of disaggregated model respectively
The threshold value of data.
12. prediction technique according to claim 11, which is characterized in that thus obtain described in future time period
Financial business of going abroad accordingly uses disaggregated model, i.e., for having used user and unused user based on the determination
People's characteristic threshold value obtains the corresponding personal characteristics data threshold in future time period by means of time sequence analysis algorithm
Value.
13. prediction technique described in 0 or 11 according to claim 1, which is characterized in that pattern recognition classifier algorithm includes random gloomy
Woods and traditional decision-tree.
14. a kind of prediction meanss (100) for the financial business potential user that goes abroad, the device include:
Acquiring unit (101), consists of, potential for obtaining the financial business of going abroad of the candidate within the time cycle occurred
The personal characteristics data of user, wherein the personal characteristics includes time series forecasting feature, invariant features and it is contemplated that feature;
Analysis and processing unit (102), consists of, and is gone abroad the potential use of financial business by the candidate obtained by the acquiring unit
The time series forecasting characteristic at family will obtain these time series forecasting features at future by time sequence analysis algorithm
Between characteristic trend curve model in the period, wherein obtaining the time series forecasting feature by the characteristic trend curve model
Corresponding characteristic data value in future time period;
Analysis and processing unit (102) is also configured as, and is gone abroad the time sequence of the prediction of financial business potential user using the candidate
Column predicted characteristics value combine corresponding invariant features value and it is contemplated that characteristic value by means of based on go abroad financial business promote user
The financial business of going abroad in corresponding future time period established predicts that the candidate goes abroad financial circles using disaggregated model
Business potential user's is that will use or will not use the financial business user that goes abroad;
Wherein the acquiring unit (101) communicates to connect via wired or wireless way and analysis and processing unit (102).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711483255.8A CN109993339A (en) | 2017-12-29 | 2017-12-29 | A kind of prediction technique for the financial business potential user that goes abroad |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711483255.8A CN109993339A (en) | 2017-12-29 | 2017-12-29 | A kind of prediction technique for the financial business potential user that goes abroad |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109993339A true CN109993339A (en) | 2019-07-09 |
Family
ID=67110549
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711483255.8A Pending CN109993339A (en) | 2017-12-29 | 2017-12-29 | A kind of prediction technique for the financial business potential user that goes abroad |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109993339A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112036959A (en) * | 2020-09-11 | 2020-12-04 | 杭州米雅信息科技有限公司 | Data processing method, device, equipment and medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101482944A (en) * | 2000-11-09 | 2009-07-15 | Spss公司 | System and method for building a time series model |
US20090292662A1 (en) * | 2008-05-26 | 2009-11-26 | Kabushiki Kaisha Toshiba | Time-series data analyzing apparatus, time-series data analyzing method, and computer program product |
CN105930875A (en) * | 2016-05-05 | 2016-09-07 | 清华大学 | User course dropping behavior prediction method and apparatus |
CN106127508A (en) * | 2016-06-17 | 2016-11-16 | 重庆大学 | A kind of Chinese crude drug price expectation method calculated based on big data |
CN106845731A (en) * | 2017-02-20 | 2017-06-13 | 重庆邮电大学 | A kind of potential renewal user based on multi-model fusion has found method |
-
2017
- 2017-12-29 CN CN201711483255.8A patent/CN109993339A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101482944A (en) * | 2000-11-09 | 2009-07-15 | Spss公司 | System and method for building a time series model |
US20090292662A1 (en) * | 2008-05-26 | 2009-11-26 | Kabushiki Kaisha Toshiba | Time-series data analyzing apparatus, time-series data analyzing method, and computer program product |
CN105930875A (en) * | 2016-05-05 | 2016-09-07 | 清华大学 | User course dropping behavior prediction method and apparatus |
CN106127508A (en) * | 2016-06-17 | 2016-11-16 | 重庆大学 | A kind of Chinese crude drug price expectation method calculated based on big data |
CN106845731A (en) * | 2017-02-20 | 2017-06-13 | 重庆邮电大学 | A kind of potential renewal user based on multi-model fusion has found method |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112036959A (en) * | 2020-09-11 | 2020-12-04 | 杭州米雅信息科技有限公司 | Data processing method, device, equipment and medium |
CN112036959B (en) * | 2020-09-11 | 2024-06-07 | 杭州米雅信息科技有限公司 | Data processing method, device, equipment and medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Giusto et al. | Identifying business cycle turning points in real time with vector quantization | |
Kali et al. | Growth networks | |
Kao et al. | A Bayesian latent variable model with classification and regression tree approach for behavior and credit scoring | |
Han et al. | Category role aided market segmentation approach to convenience store chain category management | |
CN110288484B (en) | Insurance classification user recommendation method and system based on big data platform | |
CN105931068A (en) | Cardholder consumption figure generation method and device | |
Li et al. | A financial early warning logit model and its efficiency verification approach | |
CN112561598A (en) | Customer loss prediction and retrieval method and system based on customer portrait | |
US20120109865A1 (en) | Using affinity measures with supervised classifiers | |
CN106228399A (en) | A kind of stock trader's customer risk preference categories method based on big data | |
CN110009432A (en) | A kind of personal consumption behavior prediction technique | |
CN109360022A (en) | A kind of market Sales Volume of Commodity prediction technique, device and equipment based on data mining | |
Tsai et al. | A comparative study of hybrid machine learning techniques for customer lifetime value prediction | |
CN104346698A (en) | Catering member big data analysis and checking system based on cloud computing and data mining | |
Zhou et al. | The location of new anchor stores within metropolitan areas | |
Hicham et al. | Analysis of unsupervised machine learning techniques for an efficient customer segmentation using clustering ensemble and spectral clustering | |
Wanke et al. | Revisiting camels rating system and the performance of Asean banks: a comprehensive mcdm/z-numbers approach | |
US20240087046A1 (en) | System to automatically categorize | |
CN109767333A (en) | Select based method, device, electronic equipment and computer readable storage medium | |
CN109993339A (en) | A kind of prediction technique for the financial business potential user that goes abroad | |
Ozyirmidokuz et al. | A data mining based approach to a firm's marketing channel | |
Kirshners et al. | Combining clustering and a decision tree classifier in a forecasting task | |
CN106780020A (en) | A kind of stock trend analysis and share-selecting method based on strong matching sequence | |
Atthirawong et al. | An integrated factor analysis-technique for order preference by similarity to ideal solution for location decision in ASEAN region: A case study of Thai fabric manufacturing plant | |
CN114626940A (en) | Data analysis method and device and electronic equipment |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190709 |