Closed tm-jace-peralta closed 1 year ago
Check out this pull request on
See visual diffs & provide feedback on Jupyter Notebooks.
Powered by ReviewNB
View / edit / reply to this conversation on ReviewNB
alronlam commented on 2023-02-08T03:06:49Z ----------------------------------------------------------------
Line #1. countries_data = pd.read_csv("../../data/countries_data.csv")
Jace, let's make sure the notebooks are replicable when you run from top to bottom. If I understood right, you commented out the data creation steps and instead loaded data from this file.
Understand it helps make the iteration faster, but since this is final and we're sharing it now, let's focus on replicability. Thanks!
View / edit / reply to this conversation on ReviewNB
alronlam commented on 2023-02-08T03:08:40Z ----------------------------------------------------------------
Line #20. plt.show()
Is this last part still relevant to the experiment? If not, let's just remove it!
Oh one last thing, can you add a bit of discussion at the top and refer to the 2 reference papers we found that also normalized features and wealth index per country? Just so we remember the basis for doing something like this. Thanks!
This PR adds the notebooks used to enhance cross-country model performance
2023-02-06_crosscountry_initial_jace.ipynb
- experimented on different scalers applied on the feature columns and wealth index. Performance was increased (R^2 = 0.41 -> 0.48) especially for the case where PH is the test country (R^2 = 0.25->0.42).