0

I am working on a survey data which is quarterly from 2008 to 2014 and it becomes bi-monthly 2014 onwards. Can someone please suggest how to tackle this data? It is not possible to interpolate due to many missing values? Is it appropriate to use the entire data as one single time series and then carry out my analysis?

PRASHANT MEHUL PARAB
  • 3
  • 3
  • You could combine the bimonthly data into quarters. Than all your data would be quarterly. – user54285 Mar 15 '19 at 21:48
  • @user54285! does that imply using 4 observations from the 6 given annually? – PRASHANT MEHUL PARAB Mar 17 '19 at 08:40
  • I think it would and my comment was wrong. I was thinking of monthly data. To do what I suggested you would have to take the odd bimonthly points (say the 2nd 4th 6th... point) and split the results between the quarters. – user54285 Mar 18 '19 at 16:23
  • Thanks a lot. Yesterday I was thinking of another way out. Convert bimonthly data to monthly by taking averages of the preceding and following months for missing values. Take 3 month moving averages and then take the values of every third month for analysis. Will that work out? – PRASHANT MEHUL PARAB Mar 20 '19 at 02:46
  • I am not sure. I think if there is seasonality splitting bimonthly data into months won't work period. – user54285 Mar 20 '19 at 19:35
  • But in any case I would be adjusting for the seasonality component before venturing into analysis. – PRASHANT MEHUL PARAB Apr 09 '19 at 07:27

0 Answers0