CN106997497A - A kind of bank's excess reserve Forecasting Methodology based on time series and holiday information - Google Patents
A kind of bank's excess reserve Forecasting Methodology based on time series and holiday information Download PDFInfo
- Publication number
- CN106997497A CN106997497A CN201710443928.0A CN201710443928A CN106997497A CN 106997497 A CN106997497 A CN 106997497A CN 201710443928 A CN201710443928 A CN 201710443928A CN 106997497 A CN106997497 A CN 106997497A
- Authority
- CN
- China
- Prior art keywords
- excess reserve
- holidays
- banking institution
- festivals
- statistics
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06Q40/02—Banking, e.g. interest calculation or account maintenance
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Finance (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Technology Law (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention discloses a kind of bank's excess reserve Forecasting Methodology based on time series and holiday information, based on periodicity turnover time series, holiday information is merged, banking institution's excess reserve prediction is carried out.Banking institution day deposit and day withdrawal volume historical data are changed into excess reserve time series observation data first, using the whole province's excess reserve annual average rate of increase as normative reference, banking institution's excess reserve monthly average growth rate is corrected.Then according to holiday information, independent prediction is carried out for two kinds of date properties of festivals or holidays and eekday.Finally, it is that banking institution predicts excess reserve demand using periodicity excess reserve time series observation data.This method can accurately predict most banking institution's excess reserve number, with higher precision, and can predict the provision gold number of many days.Full row cash provision management is instructed for bank, the cash demand amount of scientific forecasting mechanisms at different levels provides valuable reference information for the cash stock control of banking institution at different levels.
Description
Technical field
The invention belongs to numerical modeling prediction field, more particularly to a kind of bank based on time series and holiday information
Excess reserve Forecasting Methodology.
Background technology
Different from traditional prediction method, traditional prediction method has prediction difference larger, and degree of fitting or the degree of accuracy be not high
Feature, it is difficult to carry out actual excess reserve using predicted value and dispatch, this allows for traditional Forecasting Methodology and is difficult to meet bank's guidance
Mechanisms at different levels carry out the demand of cash library management so that bank instructs the efficiency degradation of mechanisms progress cash provisions at different levels.
The focus of bank's excess reserve scheduling Forecasting Methodology based on time series models concentrates on data in itself.In order to ensure
Premised on client's cash demand, operation operating efficiency is improved, mitigates foreground cash stock's government pressure so that mechanisms at different levels can
Complete the cash provision index that head office assigns.In the scientific method based on time series models, the cash with reference to as defined in bank
Provision volume, and using mechanism cash provision total values at different levels (being determined by history cash receipt and payment volume and history receipt and payment net amount), by two grades
Distribution principle, which is allocated, (is about to provision volume by province first to distribute to each two grades of rows, then is distributed by two grades of rows to administrative net
Point), improve the accuracy of the result of prediction, it is ensured that the cash inventory value of science, reasonable distribution mechanisms at different levels.
Bank's excess reserve scheduling Forecasting Methodology possible factor based on time series is influenceed to be, each withdrawal of client
Amount, the amount more common working day of such as festivals or holidays peak period, or even general festivals or holidays are likely to occur the spy of tens times of relative proximity
The larger situation of the number change amplitude such as different phenomenon.Macroscopically with the row or the Economic Development Status in body seat domain such as economy
Growth trend etc. is more close.
The content of the invention
The technical problems to be solved by the invention are:Based on banking institution's excess reserve have periodicity time series characteristic with
And in phenomenon of the festivals or holidays interval with larger fluctuation, it is proposed that one kind is based on time series models and festivals or holidays excess reserve information
Bank's excess reserve Forecasting Methodology, the features such as this method accuracy of identification is high, predetermined speed is fast.
The technical scheme is that:
A kind of bank's excess reserve Forecasting Methodology based on time series and holiday information, comprises the following steps:
Step 1, extracted by bank big data platform each banking institution day deposit and day withdrawal volume historical data (adopt
Collect the daily transaction journal data of bank each mechanism, and be aggregated into each mechanism day deposit and withdrawal volume), according to Day Trading total value=
Day deposit and day withdrawal volume calculate Day Trading total value and as the reference of excess reserve, build the observation of excess reserve time series
Data are simultaneously preserved to database;
Step 2, legal festivals and holidays and major event date are collectively referred to as festivals or holidays to (will be stored in festivals or holidays festivals or holidays reflects
In firing table);By t before festivals or holidays and festivals or holidays1T after it and festivals or holidays2It separately as festivals or holidays attribute statistics month,
And each calendar month excludes statistics month of the other dates outside festivals or holidays separately as a calendar month attribute;By all statistics
The moon is ranked up;If needing the excess reserve for predicting following M days, the current predictive date be the following the statistics moon where m days be y
The individual statistics moon, m=1,2 ..., M;This step proposes the concept of " the statistics moon ", can adaptive session holiday waving interval;
Step 3, the moon total transaction amount (moon total transaction amount=moon deposit-moon withdrawal volume) for calculating each banking institution, and press
Count the excess reserve monthly average growth rate that the moon calculates each banking institutionWherein, j=1,
2,…,J;J is represented comprising the banking institution at the same level for being subordinate to same upper management mechanism with it including the banking institution to be predicted
Total number, SijAnd S(i-1)jThe moon transaction of y statistics moon of j-th of expression banking institution at the same level 1 year and the i-th -1 annual control respectively
Total value, n represents the year number of data;
Step 4, annualized the whole province's excess reserve annual average rate of increaseWherein, n is
The year number of data, YiRepresent each banking institution of the whole province year total transaction amount of 1 year;
Step 5, the adjustment of excess reserve monthly average growth rate:
Step 5.1, by the whole province excess reserve annual average rate of increase V1Increase as the excess reserve monthly average of each banking institution of the whole province
Long rate V0jReference value, by V0jControl is in V1W times of stable region in;The excess reserve monthly average growth rate of each banking institution is passed through
V is obtained after adjusting for the first time2j;The V set0jLower limit MinV0j=-w* | V1|, V0jUpper limit MaxV0j=w* | V1|, exceed
This scope, then assert that its development speed is too fast for unstable development.If-w* | V1|≤V0j≤w*|V1|, then make V2j=V0jIf,
V0j<-w* | V1|, then make V2j=-w* | V1|, if otherwise V0j> w* | V1|, then make V2j=w* | V1|;Such as w values are 2.5, entirely
It is about 10% to save excess reserve annual average rate of increase, then each banking institution will be set beyond 25% excess reserve monthly average growth rate
It is that 25% or excess reserve monthly average growth rate less than -25% are arranged to -25%;
Step 5.2, first, if the time where the current predictive date is xth year, calculates the upper of the banking institution to be predicted
The total transaction amount estimate and actual value of y-th of statistics moon of level management organization's last year:
Wherein, S(x-2)jIt was that merchandising the moon for the y statistics moon of the annual control of xth -2 is total the year before last to represent j-th of banking institution at the same level
Volume, S(x-1)jIt was the moon total transaction amount of the y statistics moon of the annual control of xth -1 last year to represent j-th of mechanism at the same level;
Then, T is comparedupAnd Sup, draw coefficientAccording to monthly average growth rates of the K to the banking institution to be predicted
Carry out second of adjustment and obtain V3q, V3q=K*V3q, q ∈ { 1,2 ..., m };
Introduce equal times expanding or coefficient of reduction K so that TupAnd SupMeet condition Tup* K=Sup, draw coefficient
Wherein K > 1 represent V01Need to expand, K < 1 represent V01Need to reduce.
Step 5.3, according to the whole province excess reserve annual average rate of increase V12.5 times of stable regions again to V3qLimited,
By V3qControl in V1W times of stable region in, obtain V4q;This step is according to monthly average of the whole province's annual average rate of increase to mechanism
Growth rate carries out reasonability adjustment, adaptive correction festivals or holidays waving interval excess reserve predicted value;
Step 6, for the banking institution to be predicted, predict excess reserve according to following steps:
Step 6.1, the per day turnover for calculating y-th of statistics moon of banking institution last yearP tables
Show the number of days that y-th of statistics moon of last year is included;U is multiplied by the monthly average growth rate V of the banking institution again4q, its result is designated as Am;
Step 6.2, the attribute according to y-th of statistics moon, are predicted respectively;If the attribute of y-th of statistics moon is section
Holiday, then AmThe as excess reserve predicted value on banking institution's current predictive date;Else if the attribute of y-th of statistics moon is
Calendar month, then into step 6.3;
Step 6.3, calculate the average deal size that the banking institution goes the same period on daysIt is wherein all
Z represents all attributes on current predictive date, and all z take Monday~Sunday;N represents the number in y-th of statistics week in middle of the month z of last year;I.e.
SAll zEqual to the average deal size in y-th of statistics middle of the month all z of last year;Again by SAll zThe monthly average for being multiplied by the banking institution increases
Rate V02q, its result is designated as Bm;
Step 6.4, the banking institution all z in T days before current date average deal size is calculated Wherein L represents preceding all z in T days in addition to festivals or holidays number;T before current date
Its i.e. required number of days for extracting data, by FAll zIt is used as excess reserve average value Cm;
Step 6.5, calculating Dm=a*Bm+b*Cm+0.5*Am, it is following the m days as banking institution's current predictive date
Excess reserve predicted value, wherein m=1,2 ..., M;A and b is weight coefficient, and a+b=0.5, the value of a and b value and m has
Close, m is bigger, then a is bigger, and b is smaller.
Step 6.5 predicts the number of days M of excess reserve the need for being inputted according to user, extract the total specified number of Day Trading of nearest T days
Data are observed according to as the history of excess reserve time series;Excess reserve will be predicted by being calculated successively by step 6.1~step 6.5
The excess reserve predicted value on each date in M days, the final output excess reserve predicted value of M days.
The excess reserve demand of festivals or holidays and non-festivals or holidays can be predicted respectively by step 6, it is as a result more accurate.
Further, in the step 2, the legal festivals and holidays include New Year's Day, the Spring Festival, the Ching Ming Festival, International Labour Day, the Dragon Boat Festival
Section, the Mid-autumn Festival, National Day totally 7 legal festivals and holidays;The major event date includes college entrance examination period, admission period in spring, autumn
Admission period in season, day cat double 11 periods totally 4 major event dates;It was divided into 23 statistics moons, including 12 by 1 year accordingly
The statistics month of individual calendar month attribute, the statistics month of 11 festivals or holidays attributes.
Further, in the step 2, judge recent years each festivals or holidays it is forward and backward excess reserve time series observation
Continuously there is the number of days of singular value (crest situation) in data, so that it is determined that t1And t2Value.Before festivals or holidays and festivals or holidays
t1T after it and festivals or holidays2It is defined as festivals or holidays waving interval, and the banking institution of different scales is determined by above method respectively
Different festivals or holidays waving intervals.
Further, in the step 2, for the time of having a holiday or vacation>The longer festivals or holidays of=5 days, t1=t2=8, for other
Festivals or holidays, t1=t2=4.
Further, in the step 3, during moon total transaction amount is calculated, if the attribute of y-th of statistics moon is
Calendar month, then by single turnover exceed the statistics month the moon total transaction amount 25%, and the transaction that the history same period did not occurred
Pipelined data is filtered.
Further, in the step 5.1, w values are 2.5.
Further, in the step 6.4, T values are 28, that is, extract the Day Trading total value data of nearest 28 days (4 weeks)
Data are observed as the history of excess reserve time series.
Further, in the step 6.4, m=1 is worked as, when 2,3,4, correspondence x and y value is respectively:A=0.25,
0.3125,0.375,0.4375, b=0.25,0.1875,0.125,0.0625;
Work as m>When=5, a=0.5, b=0, i.e., now step 6.4 result of calculation CmNo longer possesses referential.
Beneficial effect:
The present invention is using day deposit and the total transaction amount of day withdrawal volume as the reference of excess reserve, in bank's big data platform
Upper acquisition banking institution history day deposit and withdrawal specified number observe data after using excess reserve time series, and it is false to merge section
Day interval excess reserve data message, defines new bank's excess reserve based on time series models and festivals or holidays excess reserve information
Forecasting Methodology, not only allow for banking institution's excess reserve has cyclophysis relation in a short time, and considers festivals or holidays area
Between excess reserve fluctuation phenomenon, independent predictions are carried out for two kinds of date properties of festivals or holidays and eekday, eventually through step 6
The excess reserve numerical value for predicting M days after the mechanism using the excess reserve time series historical data of 7*4 days before mechanism of bank.The party
Method energy success prediction goes out most excess reserve data, with higher precision and fair speed.
The method of the present invention utilizes excess reserve self adaptive amendment of the mechanism relevant information to mechanism festivals or holidays.Refer to for bank
Full row cash provision management is led, the cash demand amount of scientific forecasting mechanisms at different levels is the cash stock control of banking institution at different levels
Valuable reference information is provided.
Brief description of the drawings
Fig. 1:The flow chart of the present invention;
Fig. 2:The method compliance test result figure of the present invention;Fig. 2 (a)~Fig. 2 (c) is respectively to predict provision in different bank mechanism
Golden compliance test result figure.
Embodiment
The present invention is described in more detail below in conjunction with the drawings and specific embodiments.
Specific examples below is merely to illustrate the present invention, does not constitute limiting the scope of the invention, any not take off
From essence of the invention, the technical scheme under thinking of the present invention within protection scope of the present invention.
First, excess reserve is predicted using technical scheme disclosed in Summary;
As shown in figure 1, the invention discloses a kind of excess reserve prediction side of bank based on time series and holiday information
Method, comprises the following steps:Step 1, extracted by bank big data platform each banking institution day deposit and day withdrawal volume go through
History data, Day Trading total value is calculated and as the ginseng of excess reserve according to Day Trading total value=day deposit and day withdrawal volume
Examine, build excess reserve time series observation data and preserve to database;
Step 2, legal festivals and holidays and major event date are collectively referred to as festivals or holidays;Before festivals or holidays and festivals or holidays
4 days with 4 days after festivals or holidays (for the time of having a holiday or vacation>The longer festivals or holidays of=5 days, took before festivals or holidays 8 days and 8 days after festivals or holidays) it is single
Solely as the statistics month of a festivals or holidays attribute, and other dates outside each calendar month exclusion festivals or holidays are separately as one
The statistics month of calendar month attribute;In the present embodiment, the legal festivals and holidays include New Year's Day, the Spring Festival, the Ching Ming Festival, International Labour Day, the Dragon Boat Festival
Section, the Mid-autumn Festival, National Day totally 7 legal festivals and holidays;The major event date includes college entrance examination period, admission period in spring, autumn
Admission period in season, day cat double 11 periods totally 4 major event dates;It was divided into 23 statistics moons, including 12 by 1 year accordingly
The statistics month of individual calendar month attribute, the statistics month of 11 festivals or holidays attributes;All statistics moons are ranked up;If needing to predict not
Carry out the excess reserve of M days, the current predictive date is that the following the statistics month where m days is y-th of statistics moon, m=1,2 ..., M;This
Step proposes the concept of " the statistics moon ", can adaptive session holiday waving interval;
Step 3, the moon total transaction amount (moon total transaction amount=moon deposit-moon withdrawal volume) for calculating each banking institution, and press
Count the excess reserve monthly average growth rate that the moon calculates each banking institutionWherein, j=1,
2,…,J;J is represented comprising the banking institution at the same level for being subordinate to same upper management mechanism with it including the banking institution to be predicted
Total number, SijAnd S(i-1)jThe moon transaction of y statistics moon of j-th of expression banking institution at the same level 1 year and the i-th -1 annual control respectively
Total value, n represents the year number of data;
Step 4, annualized the whole province's excess reserve annual average rate of increaseWherein, n is
The year number of data, YiRepresent each banking institution of the whole province year total transaction amount of 1 year;
Step 5, the adjustment of excess reserve monthly average growth rate:
Step 5.1, by the whole province excess reserve annual average rate of increase V1Increase as the excess reserve monthly average of each banking institution of the whole province
Long rate V0jReference value, by V0jControl is in V12.5 times of stable regions in;The excess reserve monthly average growth rate warp of each banking institution
Cross after adjusting for the first time and obtain V2j;
Step 5.2, first, if the time where the current predictive date is xth year, calculates the upper of the banking institution to be predicted
The total transaction amount estimate and actual value of y-th of statistics moon of level management organization's last year:
Wherein, S(x-2)jIt was that merchandising the moon for the y statistics moon of the annual control of xth -2 is total the year before last to represent j-th of banking institution at the same level
Volume, S(x-1)jIt was the moon total transaction amount of the y statistics moon of the annual control of xth -1 last year to represent j-th of mechanism at the same level;
Then, T is comparedupAnd Sup, draw coefficientAccording to monthly average growth rates of the K to the banking institution to be predicted
Carry out second of adjustment and obtain V3q, V3q=K*V3q, q ∈ { 1,2 ..., m };
Step 5.3, according to the whole province excess reserve annual average rate of increase V12.5 times of stable regions again to V3qLimited,
By V3qControl in V12.5 times of stable regions in, obtain V4q;This step is put down according to the whole province's annual average rate of increase to the moon of mechanism
Equal growth rate carries out reasonability adjustment, adaptive correction festivals or holidays waving interval excess reserve predicted value;
Step 6, for the banking institution to be predicted, predict excess reserve according to following steps:
Step 6.1, the per day turnover for calculating y-th of statistics moon of banking institution last yearP tables
Show the number of days that y-th of statistics moon of last year is included;U is multiplied by the monthly average growth rate V of the banking institution again4q, its result is designated as Am;
Step 6.2, the attribute according to y-th of statistics moon, are predicted respectively;If the attribute of y-th of statistics moon is section
Holiday, then AmThe as excess reserve predicted value on banking institution's current predictive date;Else if the attribute of y-th of statistics moon is
Calendar month, then into step 6.3;
Step 6.3, calculate the average deal size that the banking institution goes the same period on daysIt is wherein all
Z represents all attributes on current predictive date, and all z take Monday~Sunday;N represents the number in y-th of statistics week in middle of the month z of last year;I.e.
SAll zEqual to the average deal size in y-th of statistics middle of the month all z of last year;Again by SAll zThe monthly average for being multiplied by the banking institution increases
Rate V02q, its result is designated as Bm;
Step 6.4, the banking institution all z in 28 days before current date average deal size is calculated Wherein L represents the number of all z in addition to festivals or holidays in first 28 days;Before current date
28 days i.e. required number of days for extracting data, by FAll zIt is used as excess reserve average value Cm;
Step 6.5, calculating Dm=a*Bm+b*Cm+0.5*Am, it is following the m days as banking institution's current predictive date
Excess reserve predicted value, wherein m=1,2 ..., M;A and b is weight coefficient, works as m=1, when 2,3,4, correspondence x and y value point
It is not:A=0.25,0.3125,0.375,0.4375, b=0.25,0.1875,0.125,0.0625;Work as m>When=5, a=
0.5, b=0, i.e., now step 6.4 result of calculation CmNo longer possesses referential.
2nd, effectiveness of the invention is verified
By the data data that 2015 are predicted as historical data before application 2015, and it is true by 2015
Real data carrys out comparison prediction data.For the accuracy (Accuracy) of more preferable statistical method, accuracy is defined as TP/ (TP+
FP), wherein TP is true positives (True Positive), and FP is false positive (False Positive).TP be defined as predicted value and
Actual value error 25% and its within (including predicted value and true error value are within 37.5 ten thousand, mainly for less number
According to the numerical value within such as 1,500,000), it is other to be defined as FP.Including the trend analysis for prediction is also counted in addition, the point is defined as
Variation tendency (rise, decline) to lower point compares, and is defined as if trend is identical very, is otherwise false.
Following table describes the excess reserve data that method predicts two grades of Hunan Province row in March, 2015 respectively and certain two grades
Row April and the excess reserve data result in May.Wherein, design sketch (a), (b) and (c) such as this explanation of specific analysis prediction
Shown in accompanying drawing 2 in book accompanying drawing.
Table 1:The inventive method predicts excess reserve accuracy table in different bank mechanism
In table 1:
Table of accuracy is shown as:Predict total points of the Expected Results for really points/prediction
Trend analysis is expressed as:The expected trend of prediction is really to count/(total points -1 of prediction);Wherein predict total points
Subtract one and be due to predicted time it is interval in last point be boundary point, it is impossible to obtain its variation tendency, therefore be not counted in trend point
In the statistics of analysis.
In addition, by using time series evaluation criterion, being estimated to the prediction effect of this method.The manner is used
Mean error (ME), mean absolute error (MAD), root-mean-square error (RMSE), mean percent ratio error (MPE), average absolute
Percentage error (MAPE) is evaluated.
Following table describes the standard meter in two grades of Hunan Province of method prediction row 2015 on January on June 9th, 1,1
Calculation value.
Table 2:The inventive method banking institution predicts excess reserve different evaluation standard value
As shown in Table 2, preferably, each error term is all smaller for this method prediction effect.
Claims (8)
1. a kind of bank's excess reserve Forecasting Methodology based on time series and holiday information, it is characterised in that including following step
Suddenly:
Step 1, the day deposit and day withdrawal volume historical data for extracting by bank big data platform each banking institution, according to day
Total transaction amount=day deposit and day withdrawal volume calculate Day Trading total value and as the reference of excess reserve, when building excess reserve
Between sequence observation data and preserve to database;
Step 2, legal festivals and holidays and major event date are collectively referred to as festivals or holidays;By t before festivals or holidays and festivals or holidays1My god
With t after festivals or holidays2It separately as festivals or holidays attribute statistics month, and each calendar month exclude it is other outside festivals or holidays
Statistics month of the date separately as a calendar month attribute;All statistics moons are ranked up;If needing to predict the standby of M days future
Fu Jin, the current predictive date is that the following the statistics where m days month is y-th of statistics moon, m=1,2 ..., M;
Step 3, the moon total transaction amount for calculating each banking institution, and increase by the excess reserve monthly average of each banking institution of statistics moon calculating
Long rateWherein, j=1,2 ..., J;J is represented comprising including the banking institution to be predicted
The banking institution's total number at the same level for being subordinate to same upper management mechanism with it, SijAnd S(i-1)jJ-th of bank at the same level is represented respectively
The moon total transaction amount of the y statistics moon of mechanism 1 year and the i-th -1 annual control, n represents the year number of data;
Step 4, annualized the whole province's excess reserve annual average rate of increaseWherein, n is data
Year number, YiRepresent each banking institution of the whole province year total transaction amount of 1 year;
Step 5, the adjustment of excess reserve monthly average growth rate:
Step 5.1, by the whole province excess reserve annual average rate of increase V1It is used as the excess reserve monthly average growth rate of each banking institution of the whole province
V0jReference value, by V0jControl is in V1W times of stable region in;The excess reserve monthly average growth rate of each banking institution passes through first
V is obtained after secondary adjustment2j;
Step 5.2, first, if the time where the current predictive date is xth year, calculates the higher level's pipe for the banking institution to be predicted
Manage the total transaction amount estimate and actual value of y-th of statistics moon of mechanism last year:
Wherein, S(x-2)jIt was the moon total transaction amount of the y statistics moon of the annual control of xth -2 the year before last to represent j-th of banking institution at the same level,
S(x-1)jIt was the moon total transaction amount of the y statistics moon of the annual control of xth -1 last year to represent j-th of mechanism at the same level;
Then, T is comparedupAnd Sup, draw coefficientThe monthly average growth rate for the banking institution to be predicted is carried out according to K
Second of adjustment obtains V3q, V3q=K*V3q, q ∈ { 1,2 ..., m };
Step 5.3, according to the whole province excess reserve annual average rate of increase V12.5 times of stable regions again to V3qLimited, by V3q
Control in V1W times of stable region in, obtain V4q;
Step 6, for the banking institution to be predicted, predict excess reserve according to following steps:
Step 6.1, the per day turnover for calculating y-th of statistics moon of banking institution last yearP represents last year
The number of days that y-th of statistics moon is included;U is multiplied by the monthly average growth rate V of the banking institution again4q, its result is designated as Am;
Step 6.2, the attribute according to y-th of statistics moon, are predicted respectively;If the attribute of y-th of statistics moon is festivals or holidays,
Then AmThe as excess reserve predicted value on banking institution's current predictive date;Else if the attribute of y-th of statistics moon is nature
Month, then into step 6.3;
Step 6.3, calculate the average deal size that the banking institution goes the same period on daysWherein week z tables
Show all attributes on current predictive date, all z take Monday~Sunday;N represents the number in y-th of statistics week in middle of the month z of last year;That is SAll z
Equal to the average deal size in y-th of statistics middle of the month all z of last year;Again by SAll zIt is multiplied by the monthly average growth rate of the banking institution
V02q, its result is designated as Bm;
Step 6.4, the banking institution all z in T days before current date average deal size is calculated Wherein L represents preceding all z in T days in addition to festivals or holidays number;T before current date
Its i.e. required number of days for extracting data, by FAll zIt is used as excess reserve average value Cm;
Step 6.5, calculating Dm=a*Bm+b*Cm+0.5*Am, it is following the m days standby as banking institution's current predictive date
Pay golden predicted value, wherein m=1,2 ..., M;A and b is weight coefficient, and a+b=0.5, a and b value and m value are relevant, m
Bigger, then a is bigger, and b is smaller.
2. bank's excess reserve Forecasting Methodology according to claim 1 based on time series and holiday information, its feature
Be, in the step 2, the legal festivals and holidays include New Year's Day, the Spring Festival, the Ching Ming Festival, International Labour Day, the Dragon Boat Festival, the Mid-autumn Festival, National Day
Save totally 7 legal festivals and holidays;The major event date includes college entrance examination period, admission period in spring, admission period in autumn, day
Totally 4 major event dates cat double 11 periods;It was divided into 23 statistics moons by 1 year accordingly, includes the system of 12 calendar month attributes
Count the moon, the statistics month of 11 festivals or holidays attributes.
3. bank's excess reserve Forecasting Methodology according to claim 1 based on time series and holiday information, its feature
Be, in the step 2, judge recent years each festivals or holidays it is forward and backward excess reserve time series observation data in it is continuous
There is the number of days of singular value, so that it is determined that t1And t2Value;T before festivals or holidays and festivals or holidays1T after it and festivals or holidays2Its definition
For festivals or holidays waving interval, the banking institution of different scales determines different festivals or holidays waving intervals by above method respectively.
4. bank's excess reserve Forecasting Methodology according to claim 3 based on time series and holiday information, its feature
It is, in the step 2, for the time of having a holiday or vacation>The longer festivals or holidays of=5 days, t1=t2=8, for other festivals or holidays, t1=
t2=4.
5. bank's excess reserve Forecasting Methodology according to claim 1 based on time series and holiday information, its feature
It is,, will if the attribute of y-th of statistics moon is calendar month during moon total transaction amount is calculated in the step 3
Single turnover exceed the statistics month the moon total transaction amount 25%, and the transaction journal data mistake that the history same period did not occurred
Filter.
6. bank's excess reserve Forecasting Methodology according to claim 1 based on time series and holiday information, its feature
It is, in the step 5.1, w values are 2.5.
7. bank's excess reserve Forecasting Methodology according to claim 1 based on time series and holiday information, its feature
It is, in the step 6.4, T values are 28, that is, when extracting the Day Trading total value data of nearest 28 days (4 weeks) as excess reserve
Between sequence history observation data.
8. bank's excess reserve Forecasting Methodology according to claim 1 based on time series and holiday information, its feature
It is, in the step 6.4, works as m=1, when 2,3,4, correspondence x and y value is respectively:A=0.25,0.3125,0.375,
0.4375, b=0.25,0.1875,0.125,0.0625;
Work as m>When=5, a=0.5, b=0, i.e., now step 6.4 result of calculation CmNo longer possesses referential.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710443928.0A CN106997497A (en) | 2017-06-13 | 2017-06-13 | A kind of bank's excess reserve Forecasting Methodology based on time series and holiday information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710443928.0A CN106997497A (en) | 2017-06-13 | 2017-06-13 | A kind of bank's excess reserve Forecasting Methodology based on time series and holiday information |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106997497A true CN106997497A (en) | 2017-08-01 |
Family
ID=59436244
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710443928.0A Pending CN106997497A (en) | 2017-06-13 | 2017-06-13 | A kind of bank's excess reserve Forecasting Methodology based on time series and holiday information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106997497A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107609974A (en) * | 2017-09-26 | 2018-01-19 | 北京福布罗科技有限公司 | Method for testing pressure and device |
CN109034493A (en) * | 2018-08-22 | 2018-12-18 | 阿里巴巴集团控股有限公司 | A kind of account balance prediction technique and device |
CN109087188A (en) * | 2018-08-21 | 2018-12-25 | 万翼科技有限公司 | Settlement method, device and the computer readable storage medium of commission |
CN109118641A (en) * | 2018-07-09 | 2019-01-01 | 北京磁云唐泉金服科技有限公司 | Accurate bank note distribution method and device based on serial number |
CN109871407A (en) * | 2019-01-23 | 2019-06-11 | 深圳壹账通智能科技有限公司 | Excess reserve data real-time report method, apparatus, computer equipment and storage medium |
CN110309399A (en) * | 2018-02-06 | 2019-10-08 | 北京嘀嘀无限科技发展有限公司 | Scene information method for pushing and device based on date classification |
CN110634061A (en) * | 2018-06-22 | 2019-12-31 | 马上消费金融股份有限公司 | Capital demand prediction method and device and terminal equipment |
CN111260156A (en) * | 2020-02-18 | 2020-06-09 | 中国农业银行股份有限公司 | Construction method of cash flow prediction model and cash flow prediction method and device |
CN111445340A (en) * | 2020-04-25 | 2020-07-24 | 中信银行股份有限公司 | Method and device for estimating value of financial product, electronic equipment and readable storage medium |
CN111639792A (en) * | 2020-05-12 | 2020-09-08 | 深圳市星耀蓝图科技有限公司 | Method for intelligently adding bank ATM (automatic teller machine) money based on artificial intelligence |
CN112101683A (en) * | 2020-09-28 | 2020-12-18 | 建信金融科技有限责任公司 | Method and device for predicting check amount of reserve money |
WO2020258994A1 (en) * | 2019-06-24 | 2020-12-30 | 深圳前海微众银行股份有限公司 | Node data prediction method and device |
CN114202292A (en) * | 2022-02-16 | 2022-03-18 | 北京每日菜场科技有限公司 | Stock information sending method and device, electronic equipment and computer readable medium |
-
2017
- 2017-06-13 CN CN201710443928.0A patent/CN106997497A/en active Pending
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107609974A (en) * | 2017-09-26 | 2018-01-19 | 北京福布罗科技有限公司 | Method for testing pressure and device |
CN110309399A (en) * | 2018-02-06 | 2019-10-08 | 北京嘀嘀无限科技发展有限公司 | Scene information method for pushing and device based on date classification |
CN110634061A (en) * | 2018-06-22 | 2019-12-31 | 马上消费金融股份有限公司 | Capital demand prediction method and device and terminal equipment |
CN109118641A (en) * | 2018-07-09 | 2019-01-01 | 北京磁云唐泉金服科技有限公司 | Accurate bank note distribution method and device based on serial number |
CN109087188A (en) * | 2018-08-21 | 2018-12-25 | 万翼科技有限公司 | Settlement method, device and the computer readable storage medium of commission |
CN109034493A (en) * | 2018-08-22 | 2018-12-18 | 阿里巴巴集团控股有限公司 | A kind of account balance prediction technique and device |
CN109871407A (en) * | 2019-01-23 | 2019-06-11 | 深圳壹账通智能科技有限公司 | Excess reserve data real-time report method, apparatus, computer equipment and storage medium |
WO2020258994A1 (en) * | 2019-06-24 | 2020-12-30 | 深圳前海微众银行股份有限公司 | Node data prediction method and device |
CN111260156A (en) * | 2020-02-18 | 2020-06-09 | 中国农业银行股份有限公司 | Construction method of cash flow prediction model and cash flow prediction method and device |
CN111260156B (en) * | 2020-02-18 | 2023-07-28 | 中国农业银行股份有限公司 | Cash flow prediction model construction method and cash flow prediction method and device |
CN111445340A (en) * | 2020-04-25 | 2020-07-24 | 中信银行股份有限公司 | Method and device for estimating value of financial product, electronic equipment and readable storage medium |
CN111639792A (en) * | 2020-05-12 | 2020-09-08 | 深圳市星耀蓝图科技有限公司 | Method for intelligently adding bank ATM (automatic teller machine) money based on artificial intelligence |
CN111639792B (en) * | 2020-05-12 | 2023-04-18 | 深圳市星耀蓝图科技有限公司 | Method for intelligently adding bank ATM (automatic teller machine) money based on artificial intelligence |
CN112101683A (en) * | 2020-09-28 | 2020-12-18 | 建信金融科技有限责任公司 | Method and device for predicting check amount of reserve money |
CN114202292A (en) * | 2022-02-16 | 2022-03-18 | 北京每日菜场科技有限公司 | Stock information sending method and device, electronic equipment and computer readable medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106997497A (en) | A kind of bank's excess reserve Forecasting Methodology based on time series and holiday information | |
Krüger et al. | The impact of loan loss provisioning on bank capital requirements | |
Kao et al. | Predicting bank performance with financial forecasts: A case of Taiwan commercial banks | |
Abdou et al. | Neural nets versus conventional techniques in credit scoring in Egyptian banking | |
Yannick et al. | Technical efficiency assessment using data envelopment analysis: an application to the banking sector of Cote d’Ivoire | |
CN107578124A (en) | The Short-Term Load Forecasting Method of GRU neutral nets is improved based on multilayer | |
CN106886846A (en) | A kind of bank outlets' excess reserve Forecasting Methodology that Recognition with Recurrent Neural Network is remembered based on shot and long term | |
CN110059896A (en) | A kind of Prediction of Stock Index method and system based on intensified learning | |
Triest | The efficiency cost of increased progressivity | |
Bhandari et al. | An artificial intelligence ATM forecasting system for hybrid neural networks | |
CN112884590A (en) | Power grid enterprise financing decision method based on machine learning algorithm | |
Nenovski | Efficiency of the Bulgarian banking system: traditional approach and data envelopment analysis | |
CN110322347A (en) | A kind of shot and long term strategy multiple-factor quantization capitalized method and device | |
CN111639792B (en) | Method for intelligently adding bank ATM (automatic teller machine) money based on artificial intelligence | |
Pavković et al. | Profitability and efficiency of the Croatian banking sector: impact of bank size | |
Noreen et al. | Model building and forecasting of bank credit to public and private sector | |
Gander | Integrating bank profit and risk-avoidance decisions for selected European countries: A micro–macro analysis | |
CN113421154B (en) | Credit risk assessment method and system based on control chart | |
Acharya et al. | Replacement hiring and the productivity-wage gap | |
CN115330147A (en) | Multi-level TOPSIS private client comprehensive evaluation method based on banking business | |
CN108615091A (en) | Electric power meteorology load data prediction technique based on cluster screening and neural network | |
CN113822751A (en) | Online loan risk prediction method | |
Ishak et al. | Utilisation of Holt-Winters Forecasting Model in Lembaga Zakat Selangor (LZS) For Zakat Collection | |
Milligan et al. | Provincial Taxation of High Incomes: The Effects on Progressivity and Tax Revenue | |
Wu et al. | User portraits and investment planning based on accounting data |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20190107 Address after: 410000 No. 678 Qingshan Road, Yuelu District, Changsha City, Hunan Province Applicant after: Hunan Creator Information Technologies Co.,Ltd. Address before: 410083 932 south foot Road, Yuelu District, Changsha, Hunan Applicant before: Central South University Applicant before: Hunan Hyderabad Data Technology Co. Ltd. |
|
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170801 |