1

I am new to time series regression. I am trying to understand the effect of differencing the time-dependent data before applying regression analysis.

I tried fitting lots of regression models but in all those cases the model assumptions are violated and show a high autocorrelation.

I was told that interpretation of the model is extremely hard when differencing is applied. But my objective is to determine the variable importance; therefore I want to understand whether applying differencing has an effect on variable importance.

What other methods can I use to understand variable importance in time series variable?

User1865345
  • 8,202
Nipuni Opatha
  • 57
  • 1
    Hi Nipuni. Welcome to CV. Could you elucidate which model assumptions you are talking about? What is the motivation of (not) differencing before applying regression? Any elaboration would be appreciated. – User1865345 Jun 27 '23 at 12:20
  • About the interpretation: https://stats.stackexchange.com/questions/74924/interpreting-coefficients-of-first-differences-of-logarithms – kk68 Jun 27 '23 at 12:32
  • Hello, So when I applied regression analysis on the data, Normality assumption, multicollinearity, and all residual analysis assumptions were violated. I tried applying transformations like log transformations and square root transformations but still, the models did not satisfy the assumptions. In all the models, residuals showed a high autocorrelation. My teacher suggested using a variable selection method, so I applied the backward selection method and also tried all possible models method but still a satisfying model cannot be found because of the autocorrelation. – Nipuni Opatha Jun 27 '23 at 15:09
  • There's a trend pattern in the data, therefore I used differencing on the data and then applied regression then the model assumptions were all satisfied. But I'm not sure if that is a valid, meaningful method. – Nipuni Opatha Jun 27 '23 at 15:09
  • @NipuniOpatha, unless we get a glimpse of your data, it is too speculative to give a substantial answer. Meanwhile, you can check the site for related posts like this, this, this, (ctd.) – User1865345 Jun 28 '23 at 03:11
  • this, this, this etc. – User1865345 Jun 28 '23 at 03:12
  • 1
    @User1865345-solidarityMods, Thank you very much for the help. I will look up those posts. – Nipuni Opatha Jun 28 '23 at 08:24

0 Answers0