SSPACE persistence modelling
Posted: Thu Aug 08, 2013 12:17 pm
Dear forum,
I'm a student and a food company asked me to analyse their 100 products on the possibility of persistent promotion effects. Now I'm trying to estimate state space models in a program I have never used before: Eviews. I'm very accustomed to Stata but that program doesn't have state space models with the possibility of time-varying coefficients.
The syntax I have written is wrong according to the program and I have to acknowledge that after many trials I need help: also on the topics of state space models. To make things troublesome: It needs to be done in one week and the professor who knows a lot about the topic isn't available in that time period. To be specific; I'm sure the model itself is good but I don't know much about the syntax.
Since there may be experts out there who can help in just a minute of their time I would like to ask them to help me out with the syntax. Here is a simplification of the syntax (a local linear trend model):
---------------------------------------------------------------------------------------------
'sspace model (verkoop=sales and televisiebestedingen=promotion via television)
series y=verkoop
series x=televisiebestedingen
sspace ss1
ss1.append @signal y= sv1 + sv2*x+c(1)*x(-1)+[var = exp(c(2))]
ss1.append @state sv1 = sv1(-1)+sv3(-1)*x+[var = exp(c(3))]
ss1.append @state sv3 = sv3(-1)+sv4(-1)+[var = exp(c(4))]
ss1.append @state sv4 = sv4(-1)+[var = exp(c(5))]
ss1.ml(showopts,m=500,c=1e-06)
---------------------------------------------------------------------------------------------
How can I estimate this model, get coeficients c(1) etc. and the state variables? What syntax is missing or wrong? According to the program there's something wrong with ''ss1.append @signal y= sv1 + sv2*x+c(1)*x(-1)+[var = exp(c(2))]'' and "ss1.ml(showopts,m=500,c=1e-06)".
I'm trying to find the solution myself but any help is very appreciated! I have attached the simplified dataset.
Thank you in advance!
I'm a student and a food company asked me to analyse their 100 products on the possibility of persistent promotion effects. Now I'm trying to estimate state space models in a program I have never used before: Eviews. I'm very accustomed to Stata but that program doesn't have state space models with the possibility of time-varying coefficients.
The syntax I have written is wrong according to the program and I have to acknowledge that after many trials I need help: also on the topics of state space models. To make things troublesome: It needs to be done in one week and the professor who knows a lot about the topic isn't available in that time period. To be specific; I'm sure the model itself is good but I don't know much about the syntax.
Since there may be experts out there who can help in just a minute of their time I would like to ask them to help me out with the syntax. Here is a simplification of the syntax (a local linear trend model):
---------------------------------------------------------------------------------------------
'sspace model (verkoop=sales and televisiebestedingen=promotion via television)
series y=verkoop
series x=televisiebestedingen
sspace ss1
ss1.append @signal y= sv1 + sv2*x+c(1)*x(-1)+[var = exp(c(2))]
ss1.append @state sv1 = sv1(-1)+sv3(-1)*x+[var = exp(c(3))]
ss1.append @state sv3 = sv3(-1)+sv4(-1)+[var = exp(c(4))]
ss1.append @state sv4 = sv4(-1)+[var = exp(c(5))]
ss1.ml(showopts,m=500,c=1e-06)
---------------------------------------------------------------------------------------------
How can I estimate this model, get coeficients c(1) etc. and the state variables? What syntax is missing or wrong? According to the program there's something wrong with ''ss1.append @signal y= sv1 + sv2*x+c(1)*x(-1)+[var = exp(c(2))]'' and "ss1.ml(showopts,m=500,c=1e-06)".
I'm trying to find the solution myself but any help is very appreciated! I have attached the simplified dataset.
Thank you in advance!