Page 1 of 1

variance nonstationary

Posted: Thu Sep 01, 2011 11:17 am
by des_poi_na1988
hellow!! my topic is: analysis of the production of Japanese cars in the period: 1949-1988. I can not understand if my time serie is nonstationary in the variance from the line graph! :( i must take the logarithmic transformation of the series? i would appreciate if you help me. :) thanks!!!
line graph.doc
(30 KiB) Downloaded 121 times

Re: variance nonstationary

Posted: Sun Sep 11, 2011 10:17 pm
by isaiah
I don't think your series look stationary

Re: variance nonstationary

Posted: Mon Sep 12, 2011 8:23 am
by dimitris
u should the adf test or PP to have more accurate results/ also u can see the stationarity from the correlogram . not only the graph. but by the graph it seems stationary

Re: variance nonstationary

Posted: Mon Sep 12, 2011 9:14 am
by dimitris
coorect : it seems stationary

R squared

Posted: Mon Oct 24, 2011 10:16 pm
by isaiah
Dear All,

I am just estimating regression equation using pooled OLS, I have 13 variables and 180 observation. I have added one variable at a time and when I had added 10 variables, my R squared was 0.77 and when I tried to add the elevent, my R dropped to 0.63 . I tested the autocorrelation and heteroscadasicity and seems to be no problem. Why is R squared dropping? Shall I drop the variables from my model?


Re: variance nonstationary

Posted: Tue Oct 25, 2011 6:36 am
by startz
Check to see if your sample size changed.