Regression Analysis with Time Series Data

#1
Hey guys, I have some monthly time series data and I'm trying to perform regression analysis on it. I'm just wondering, if I have a few insignificant months does it make sense to remove their parameters from the model? It marginally increases my adjusted R-squared, but I'm just wondering if it'll mess up my prediction if I've gotten rid of a few months.

Thanks!
 

noetsi

No cake for spunky
#2
Well first off linear or logistic regression won't work with time series data if there is any autocorrelation as there commonly is. None of the time series methods I know [ARIMA, exponential smoothing, Regression with autoregressive error etc] allow you to remove months.