I have some observed data (like the household income, expenditure, etc.) which are not in annually (like 1996, 2001, 2005... and so on). I want to make annual data series from the available observed data. So can anyone suggest me a method to fill those gaps with the available limited number of data?
Thanking you in advance
Perhaps you might find other data from each year, and look at cross-sectional- like regression ('prediction,' not forecasting) with a data point from each year. That would make each year that you have the y-value a complete data point. The difference between years would be taken into consideration by the different x-values for each regressor (if you can find such regressors), for each data point. Assuming the x-values are impacted by time the same way the y-values are, you may be ok. Once you have filled in these gaps, perhaps you can then see if the y value time series makes sense in each case. For some regressor data, you might look at administrative data. - I'm just guessing, and you may not find appropriate regressor data. - The estimated variance of the prediction error would always be important. - Cheers.
Dear De Silva
Yes, regression can be a predictive technique for the case, however you will be needing some independent data for the same, which I am sure, would not be easily available.
I recommend to use Interpolation Method. there are numerous Interpolation Methods for different situation. You read the property of these methods and select the suitable one. Yes, you have to mention this in your document.
REgards
Rajiv
Yes, interpolation for the in-between missing data points and extrapolation for the missing data at the very beginning and end. This practice is common in traffic engineering applications and studies. As mentioned above, this step must be reported.
Dear Professors,
Thank you very much for your answers. I'll try with regression analysis technique.
Regards,
Gouri
- Badges
- Science topic