Additive model and simulation of the time series

Question

panik772 illza 2015-1-4

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/168895-additive-model-and-simulation-of-the-time-series

Hello,

It will be great if somebody could help me with following task.. Im trying to solve it about 3 days and read a lot of papers according this issue, but it doesnt work

My task is to simulate a time series with Monte Carlo method.. and Im trying to solve in this way

The time series calls A So, at first i want to fit there to additive component model and estimate the parameters in Matlab

"X=m+s+y(t) " where " m=a1+b1*t " and " sum(m-x)ˆ2=min " (least square approach)

 in this case m is a trend model with parameters a1 and b1 
  "s "   is  s a saisonal model  with parameters c1 and c2  
and  y(t) is  stochastic part, which determinate as AR(1)  modeland so on .

how can I estimate this parameters (a1, b1, c1, ) Im trying to solve it with funtion nlintool(x, y, @myfun, b0) but is it a right way ?may be there is a better tool to estimate the parameters, or common used toolbox in Matlab according to additive model and time series analyse ?

and then i need to simulate this time series with estimated model parameters in Monte Carlo simulation. .can I use direct simulation without parameter estimation in any toolboxes? is it neccessary to fit the model first? and which tool one can use according this issue thank your very much