Refactoring of regression estimates#465
Conversation
Names of regressors are synchronised with those used in Simpaths ; new master file with conditions is created
|
@dav-sonn @justin-ven Could you please review this? |
The integration and run tests are failing because the regressor names need to be updated within the code as well. @justin-ven, please let me know how you would like to proceed. I can update the code and push the changes to this PR, then leave the final review to you. |
Sounds sensible to me David |
Thanks for the confirmation, Justin! I should have this done by Monday. I will drop you a message when it is ready to be reviewed. |
Update Person.java to add new enum variables and corresponding getDoubleValue cases according to the new regressor names. Add InvMillsRatio as an alias to InverseMillsRatio. These changes restore compatibility with refactored estimation files. Also update input/reg_education.xlsx and input/reg_health.xlsx.
…rmed variable In PR #465, Year_transformed was simply renamed demYear. With no specification in the name, demYear was simply added without the transformation, which skewed the magnitude of the coefficients. This bug was corrected in this commit.
3 participants
Hi,
This version includes the following changes:
(1) Incorporated Stata programs (created by Ashley) and refactored names of regressors across all processes, except for mental health, well-being, and financial distress. I followed the new codebook naming conventions. However, for some dummies crated from categorical variables the resulting names exceeded Stata’s 32 character variable name limit, so I shortened them where necessary in a reasonably intuitive way.
(2) Moved all conditions into a new do-file called "00_master_conditions" (as in the EU version). This file is now called from the master do-file for regression estimates. It should also be integrated into the initial populations do-files at some point.
Daria