CN103399867B - A kind of self-adapting regulation method of linear combination prediction model weight - Google Patents
A kind of self-adapting regulation method of linear combination prediction model weight Download PDFInfo
- Publication number
- CN103399867B CN103399867B CN201310283242.1A CN201310283242A CN103399867B CN 103399867 B CN103399867 B CN 103399867B CN 201310283242 A CN201310283242 A CN 201310283242A CN 103399867 B CN103399867 B CN 103399867B
- Authority
- CN
- China
- Prior art keywords
- prime
- weight
- delta
- sigma
- actual result
- 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.)
- Active
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of self-adapting regulation method of linear combination prediction model weight, it is characterized in that: (1) weight w
i,mwith forecast model f
ithe m time outer before the result of calculation of k-1 time relevant to front k actual result related coefficient; (2) weight w
i,mwith forecast model f
ithe m time result of calculation relevant with the deviation of actual result, the forecast model less with nearest actual result deviation, arranges higher weight; (3) what calculate that (1) related coefficient of obtaining and calculating (2) obtains respectively predicts the outcome and the deviation of actual result, thus reaches the object calculating adaptive weighting.The present invention has fully taken into account and has predicted the outcome with the deviation of actual result and predict the outcome and the related coefficient of front k result in the weight self-adaptative adjustment of linear combination model calculates, and adds accuracy and the rationality of algorithm.
Description
Technical field
The present invention relates to a kind of in tax index prediction, predicting the outcome and actual result according to linear combination forecasting model, dynamically changes the method for the weight of each forecast model.
Background technology
In combination linear prediction, each forecast model role in prediction is not changeless, but affects dynamic change by all kinds of extraneous factor.Such as, in tax index prediction, each forecast model is by the impact of macroeconomy situation, all kinds of economy and the tax policy.Because all kinds of factor is to the difference of forecast model influence degree, the weight of each forecast model, also in dynamic change, in order to ensure the performance of linear combination forecasting model, needs to carry out dynamic conditioning adaptively to it.Applicant is new through looking into, and retrieved one section of relevant patent: weight adjusting module and weight regulating method [application publication number: CN1925543].In that patent, inventor provides a kind of weight adjusting module and weight regulating method, is applicable to adjust the weight in image-zooming technology.It is all for embody rule that above-mentioned Patents invents described method, and algorithm idea and the self-adaptative adjustment Weight algorithm that this method adopts have essence different.
Summary of the invention
The object of this invention is to provide one can according to predicted value and actual result, dynamic alignment combination forecasting
in i-th forecast model f
iweights omega
i(i ∈ [1..n]) carries out the method for accommodation.
For reaching above object, the present invention takes following technical scheme to be achieved:
A self-adapting regulation method for linear combination prediction model weight, is characterized in that, in accordance with the following steps:
(1) initial weight is determined: set the numerical value of actual result first as F '
0, each forecast model f
ipredicted value be F '
i; For each i ∈ [1..n], calculate
then each forecast model f
iinitial weight be
meet the model that prediction effect is good first and there is higher initial weight, and
(2) dynamic conditioning weight:
A, establish current totally to predict the outcome for m time and actual result, comprise the F ' that predicts the outcome of before except the m time k-1 time
i, m-k+1, F '
i, m-k+2.., F '
i, m-1, and the actual result F ' of first k time
m-k+1,
F '
m-k+2.., F '
m; Wherein, k is time window, defaultly gets 5; As m < k, get k=m; Corresponding F '
i, m-k+1, F '
i, m-k+2.., F '
i, m-1weight be respectively w
i, m-k+1, w
i, m-k+2.., w
i, m-1;
B, forecast model f
i, i ∈ [1..n], the result of calculation of k-1 time and the related coefficient of front k actual result before except the m time;
C, forecast model f
ithe deviation of the m time result of calculation and actual result,
D, result according to A, B, self-adaptation solves each forecast model f
ithe weights omega of (i ∈ [1..n]) the m time adjustment
i,m, for the forecast model high and less with actual result deviation with actual result correlativity, increase the value of weight; For the forecast model low with actual result correlativity and the forecast model larger with actual result deviation, reduce the value of weight.
In such scheme, in described step (2), the specific algorithm of step B is:
STEP1. calculate
In order to avoid weight adjusting too frequently causes shake, as F '
m=F
m=0 or | F '
m-F
m|/(| F '
m|+| F
m|)≤5%, do not carry out index weights adjustment, namely for each i ∈ [1..n], make ω
i,m=ω
i, m-1, return (ω
1, m, ω
2, m.., ω
i,m.., ω
n,m), algorithm exits;
STEP2. according to forecast model f
ithe result of calculation of (i ∈ [1..n]) and the related coefficient of actual result, determine weight adjusting coefficient:
STEP2.1. for each i ∈ [1..n], Pearson product-moment correlation coefficient (PPMCC) sequence of calculation F ' is adopted
i, m-k+1, F '
i, m-k+2.., F '
i, m-1with F '
m-k+1, F '
m-k+2.., F '
m-1related coefficient, obtain one group of correlation coefficient r
i, r
i∈ [-1,1];
In formula (1),
STEP2.2. to correlation coefficient r
i(i ∈ [1..n]) carries out following zero correction, obtains one group of weight adjusting coefficient r '
i, r '
ishow more greatly forecast model f
ihigher with actual result correlativity;
In described step (2), the specific algorithm of step C is:
Computational prediction model f
i(i ∈ [1..n]) the m time result of calculation F '
i,mwith actual result F '
mdeviation Δ F
i:
Deviation Δ F
ilarger, show forecast model f
ithe m time forecasting accuracy high.
In described step (2), the specific algorithm of D step is:
STEP1. computational prediction model f
ithe weight adjusting value Δ ω of (i ∈ [1..n])
i,m:
STEP1.1. to correlation coefficient r '
i(i ∈ [1..n]) and deviation Δ F
ispan normalization:
STEP1.2. computational prediction model f
ithe weight adjusting value Δ ω of (i ∈ [1..n])
i,m
In formula (7), p
1and p
2be respectively correlation coefficient r '
i, deviation Δ F
iweight, wherein p
1+ p
2=1, p under default setting
1=0.4p
2=0.6;
STEP2. for each i ∈ [1..n], ω is calculated
i,m;
ω′
i,m=ω
i,m-1+Δω
i,mi∈[1..n] (8)
Weight adjusting value Δ ω
i,mmeet
return (ω
1, m, ω
2, m.., ω
i,m.., ω
n,m).
Advantage of the present invention is, make use of forecast model f
ithe m time outer before the result of calculation of k-1 time and front k actual result related coefficient and forecast model f
ithe m time result of calculation and the deviation of actual result add the accuracy rate and rationality that algorithm calculates.
Accompanying drawing explanation
Below in conjunction with the drawings and the specific embodiments, the present invention is described in further detail.
Fig. 1 the inventive method process flow diagram.
Embodiment
What the present invention adopted is predicting the outcome and actual result according to linear combination forecasting model, dynamically changes the method for the weight of each forecast model.
Research purpose: predicting the outcome and actual result according to linear combination forecasting model, dynamically changes the weight of each tax index model.
Research background: in combination linear prediction, each forecast model role in prediction is not changeless, but affects dynamic change by all kinds of extraneous factor.Such as, in tax index prediction, each forecast model is by the impact of macroeconomy situation, all kinds of economy and the tax policy.Because all kinds of factor is to the difference of forecast model influence degree, the weight of each forecast model, also in dynamic change, in order to ensure the performance of linear combination forecasting model, needs to carry out dynamic conditioning adaptively to it.
As shown in Figure 1, a kind of self-adapting regulation method of linear combination prediction model weight, comprise 6 steps, its idiographic flow is:
(1) be F ' according to the numerical value of actual result first
0with each forecast model f
ipredicted value be F '
i, for each i ∈ [1..n], calculate each forecast model f
iinitial weight w
i, 0;
(2) in order to avoid weight adjusting too frequently causes shake, if F '
m=Fm=0 or | F '
m-F
m|/(| F '
m|+| F
m|)≤5%, do not carry out index weights adjustment, make ω
i,m=ω
i, m-1, algorithm exits, otherwise performs (3).
(3) according to forecast model f
ithe result of calculation of (i ∈ [1..n]) and the related coefficient of actual result, determine weight adjusting coefficient:
(4) computational prediction model f
i(i ∈ [1..n]) the m time result of calculation F '
i,mwith actual result F '
mrelative deviation Δ F
i:
(5) computational prediction model f
ithe weight adjusting value Δ ω of (i ∈ [1..n])
i,m;
(6) for each i ∈ [1..n], ω is calculated
i, m=ω
i, m-1+ Δ ω
i,m; Return (ω
1, m, ω
2, m.., ω
i,m.., ω
n,m), algorithm exits.
According to following detailed step:
(1) initial weight is determined: set the numerical value of actual result (actual value of tax index) first as F '
0, each forecast model f
ipredicted value be F '
i; For each i ∈ [1..n], calculate
Then each forecast model f
iinitial weight be
Meet the model that prediction effect is good first and there is higher initial weight, and
(2) dynamic conditioning weight: establish current totally to predict the outcome for m time and actual result; For each i ∈ [1..n], according to forecast model f
ithe result of calculation of k-1 time and front k actual result related coefficient before except the m time, and forecast model f
ithe deviation of the m time result of calculation and actual result, solves each forecast model f
ithe weights omega of (i ∈ [1..n]) the m time adjustment
i,m; For the forecast model high and less with nearest actual result deviation with actual result correlativity in early stage, increase the value of weight; For the forecast model low with actual result correlativity and the forecast model larger with nearest actual result deviation, reduce the value of weight.
Described dynamic conditioning weight performs according to following algorithm:
Input:
A) forecast model f
i(i ∈ [1..n]) except the m time before k-1 time (k is time window, defaultly gets 5; As m < k, get k=m) result of calculation be F '
i, m-k+1, R '
i, m-k+2.., F '
i, m-1;
B) corresponding F '
i, m-k+1, F '
i, m-k+2.., F '
i, m-1weight be respectively w
i, m-k+1, w
i, m-k+2.., w
i, m-1;
C) the actual result F ' of first k time
m-k+1, F '
m-k+2.., F '
m;
Export: vector (ω
1, m, ω
2, m.., ω
i,m.., ω
n,m), ω
i,mfor forecast model f
ithe weight adjusting result of the m time;
Step is:
STEP1. calculate
In order to avoid weight adjusting too frequently causes shake, as F '
m=F
m=0 or | F '
m-F
m|/(| F '
m|+| F
m|)≤5%, do not carry out index weights adjustment, namely for each i ∈ [1..n], make ω
i,m=ω
i, m-1, return (ω
1, m, ω
2, m.., ω
i,m.., ω
n,m), algorithm exits;
STEP2. according to forecast model f
ithe result of calculation of (i ∈ [1..n]) and the related coefficient of actual result, determine weight adjusting coefficient:
STEP2.1. for each i ∈ [1..n], Pearson product-moment correlation coefficient (PPMCC) sequence of calculation F ' is adopted
i, m-k+1, F '
i, m-k+2.., F '
i, m-1with F '
m-k+1, F '
m-k+2.., F '
m-1related coefficient, obtain one group of correlation coefficient r
i, r
i∈ [-1,1];
In formula (1),
STEP2.2. to correlation coefficient r
i(i ∈ [1..n]) carries out following zero correction, obtains one group of weight adjusting coefficient r '
i, r '
ishow more greatly forecast model f
ihigher with actual result correlativity;
STEP3. computational prediction model f
i(i ∈ [1..n]) the m time result of calculation F '
i,mwith actual result F '
mdeviation Δ F
i:
STEP3.1.
STEP3.2.
Deviation Δ F
ilarger, show forecast model f
ithe m time forecasting accuracy high.
STEP4. computational prediction model f
ithe weight adjusting value Δ ω of (i ∈ [1..n])
i,m:
STEP4.1. to correlation coefficient r '
i(i ∈ [1..n]) and deviation Δ F
ispan normalization:
STEP4.2. computational prediction model f
ithe weight adjusting value Δ ω of (i ∈ [1..n])
i,m
In formula (7), p
1and p
2be respectively correlation coefficient r '
i, deviation Δ F
iweight, wherein p
1+ p
2=1 (p under default setting
1=0.4p
2=0.6).
STEP5. for each i ∈ [1..n], ω is calculated
i,m;
ω′
i,m=ω
i,m-1+Δω
i,mi∈[1..n] (8)
Weight adjusting value Δ ω
i,mmeet
return (ω
1, m, ω
2, m.., ω
i,m.., ω
n,m), algorithm exits.
Claims (1)
1. a self-adapting regulation method for linear combination prediction model weight, is characterized in that, in accordance with the following steps:
(1) initial weight is determined: set the numerical value of actual result first as F '
0, each forecast model f
ipredicted value be F '
i, wherein i ∈ [1...n], calculates
then the initial weight of each forecast model fi is
meet the model that prediction effect is good first and there is higher initial weight, and
(2) dynamic conditioning weight:
A, establish current totally to predict the outcome for m time and actual result, comprise the F ' that predicts the outcome of before except the m time k-1 time
i, m-k+1, F '
i, m-k+2..., F '
i, m-1, and the actual result F ' of first k time
m-k+1, F '
m-k+2..., F '
m; Wherein, k is time window, defaultly gets 5; As m < k, get k=m; Corresponding F '
i, m-k+1, F '
i, m-k+2..., F '
i, m-1weight be respectively ω
i, m-k+1, ω
i, m-k+2..., ω
i, m-1;
B, forecast model f
i, i ∈ [1...n], the result of calculation of k-1 time and the related coefficient of front k actual result before except the m time;
C, computational prediction model f
ithe deviation of the m time result of calculation and actual result,
D, result according to A, B, self-adaptation solves each forecast model f
ithe weights omega of the m time adjustment
i, m, wherein i ∈ [1...n], for the forecast model high and less with actual result deviation with actual result correlativity, increases the value of weight; For the forecast model low with actual result correlativity and the forecast model larger with actual result deviation, reduce the value of weight;
In described step (2), the specific algorithm of step B is:
STEP1. calculate
In order to avoid weight adjusting too frequently causes shake, as F '
m=F
m=0 or | F '
m-F
m|/(| F '
m|+| F
m|)≤5%, do not carry out index weights adjustment, namely for each i ∈ [1...n], make ω
i, m=ω
i, m-1, return (ω
1, m, ω
2, m..., ω
i, m..., ω
n, m), algorithm exits;
STEP2. according to forecast model f
iresult of calculation and the related coefficient of actual result, determine weight adjusting coefficient:
STEP2.1. for each i ∈ [1...n], Pearson product-moment correlation coefficient PPMCC sequence of calculation F ' is adopted
i, m-k+1, F '
i, m-k+2..., F '
i, m-1with F '
m-k+1, F '
m-k+2..., F '
m-1related coefficient, obtain one group of correlation coefficient r
i, r
i∈ [-1,1];
In formula (1),
STEP2.2. to correlation coefficient r
icarry out following zero correction, obtain one group of weight adjusting coefficient r '
i, i ∈ [1...n], r '
ishow more greatly forecast model f
ihigher with actual result correlativity;
In described step (2), the specific algorithm of step C is:
Computational prediction model f
ithe m time result of calculation F '
i, mwith actual result F '
mdeviation Δ F
i:
Deviation Δ F
ilarger, show forecast model f
ithe m time forecasting accuracy high;
In described step (2), the specific algorithm of D step is:
STEP1. computational prediction model f
iweight adjusting value Δ ω
i, m:
STEP1.1. to correlation coefficient r
iwith deviation Δ F
ispan normalization:
Wherein: r '
ifor r
inormalized value, Δ F '
ifor Δ F
inormalized value;
STEP1.2. computational prediction model f
iweight adjusting value Δ ω
i, m
In formula (7), p
1be correlation coefficient r '
iweight, p
2for deviation Δ F
iweight, wherein p
1+ p
2=1, p under default setting
1=0.4, p
2=0.6;
STEP2. for each i ∈ [1...n], ω is calculated
i, m;
ω′
i,m=ω
i,m-1+Δω
i,mi∈[1...n] (8)
Weight adjusting value Δ ω
i, mmeet
return (ω
1, m, ω
2, m..., ω
i, m..., ω
n, m).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310283242.1A CN103399867B (en) | 2013-07-05 | 2013-07-05 | A kind of self-adapting regulation method of linear combination prediction model weight |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310283242.1A CN103399867B (en) | 2013-07-05 | 2013-07-05 | A kind of self-adapting regulation method of linear combination prediction model weight |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103399867A CN103399867A (en) | 2013-11-20 |
CN103399867B true CN103399867B (en) | 2015-10-28 |
Family
ID=49563497
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310283242.1A Active CN103399867B (en) | 2013-07-05 | 2013-07-05 | A kind of self-adapting regulation method of linear combination prediction model weight |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103399867B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105608317B (en) * | 2015-12-18 | 2018-06-26 | 上海集成电路研发中心有限公司 | A kind of digital filter apparatus and method based on linear system |
CN109284437B (en) * | 2018-08-01 | 2020-12-08 | 广东奥博信息产业股份有限公司 | Weight adaptive feedback adjustment method and device for information push |
CN116760033B (en) * | 2023-08-21 | 2024-04-12 | 南京博网软件科技有限公司 | Real-time power demand prediction system based on artificial intelligence |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101695050A (en) * | 2009-10-19 | 2010-04-14 | 浪潮电子信息产业股份有限公司 | Dynamic load balancing method based on self-adapting prediction of network flow |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8355938B2 (en) * | 2006-01-05 | 2013-01-15 | Wells Fargo Bank, N.A. | Capacity management index system and method |
-
2013
- 2013-07-05 CN CN201310283242.1A patent/CN103399867B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101695050A (en) * | 2009-10-19 | 2010-04-14 | 浪潮电子信息产业股份有限公司 | Dynamic load balancing method based on self-adapting prediction of network flow |
Non-Patent Citations (2)
Title |
---|
一种新的非线性模糊自适应变权重组合预测模型;鄂加强等;《模糊系统与数学》;20060831;第20卷(第4期);第123页-第127页 * |
一种求解组合预测模型权重的新方法;王吉权等;《数学的实践与认识》;20080131;第38卷(第1期);第76页-第81页 * |
Also Published As
Publication number | Publication date |
---|---|
CN103399867A (en) | 2013-11-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107120721B (en) | A kind of central heating dynamic gas candidate compensation method | |
US11326579B2 (en) | Adaptive dynamic planning control method and system for energy storage station, and storage medium | |
Duan et al. | Forecasting Crude Oil Consumption in China Using a Grey Prediction Model with an Optimal Fractional‐Order Accumulating Operator | |
CN103399867B (en) | A kind of self-adapting regulation method of linear combination prediction model weight | |
CN103490956A (en) | Self-adaptive energy-saving control method, device and system based on traffic predication | |
CN105719028B (en) | A kind of air conditioner load dynamic prediction method based on multifactor chaos support vector machines | |
CN106485339A (en) | A kind of part throttle characteristics of power system determines method and system | |
CN102968055A (en) | Fuzzy PID (Proportion Integration Differentiation) controller based on genetic algorithm and control method thereof | |
CN103336891B (en) | A kind of pseudo-measurement generation method estimated for state of electric distribution network | |
CN105631528B (en) | Multi-target dynamic optimal power flow solving method based on NSGA-II and approximate dynamic programming | |
CN107577876A (en) | A kind of spiral bevel gear flank of tooth loading performance Multipurpose Optimal Method | |
CN106971548B (en) | The Short-time Traffic Flow Forecasting Methods of optimizable multi-core adaptive support vector machines | |
CN109409614A (en) | A kind of Methods of electric load forecasting based on BR neural network | |
CN104598765A (en) | Building energy consumption prediction method based on elastic adaptive neural network | |
CN108090307B (en) | Multi-working-condition plate-fin heat exchanger channel layout design method based on integral average temperature difference method | |
CN103853918A (en) | Cloud computing server dispatching method based on idle time prediction | |
CN109147324B (en) | Traffic jam probability forecasting method based on user feedback mechanism | |
CN109088742A (en) | A kind of traffic forecast method and network element device, computer readable storage medium | |
CN114417631A (en) | Irrigation area water transmission and distribution system modeling method based on observation data | |
CN106468467A (en) | A kind of air-conditioning refrigeration duty real-time estimate algorithm being applied to embedded control system | |
CN116470511A (en) | Circuit power flow control method based on deep reinforcement learning | |
Xie et al. | A novel mutual fractional grey bernoulli model with differential evolution algorithm and its application in education investment forecasting in China | |
CN104063810B (en) | A kind of electricity consumption factor based on big data influences the construction method of model | |
CN105389625A (en) | Active distribution network ultra-short term load prediction method | |
CN102158164B (en) | Trapezoidal variable weight regression control method for magnetically controlled shunt reactor |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
C41 | Transfer of patent application or patent right or utility model | ||
TR01 | Transfer of patent right |
Effective date of registration: 20160415 Address after: 310053, tax building, No. 3738 South Ring Road, Hangzhou, Zhejiang, Binjiang District Patentee after: Servyou Software Group Co., Ltd. Address before: 710049 Xianning West Road, Shaanxi, China, No. 28, No. Patentee before: Xi'an Jiaotong University |