Dalia Ghanem - Model Selection and prediction after selection in linear fixed effects models with high-frequency regressors


243
views
0
12 months ago by
Ghanem Slides (click here)
Community: WHOA-PSI 2017

2 Answers


0
12 months ago by
Have you considered modeling using wavelets rather than bins?
This is definitely much better than what we have done so far.  We have been motivated by what empirical researchers are doing and keeping out method close to what they might do, but I believe looking at wavelets would be a very fruitful step.  Thank you!
written 12 months ago by Dalia Ghanem  
0
12 months ago by
Do you consider  temporal-spatial modeling for this data? It would be interesting to test dynamic model of time.
This is a great point! Practitioners tend to allow for spatial dependence and resample time periods as if they were independent.  Of course, ignoring the time series dependence.  I believe this is a very important direction to take the theoretical results, namely to come up with a cross-validation procedure that allow for both time series and spatial correlation.  Thank you for the comment!
written 12 months ago by Dalia Ghanem  
Please login to add an answer/comment or follow this question.