Midterm report review

[mchao409]

This project is about using voter demographics in swing state counties to predict if they will lean red or blue in the 2020 election. There are 12 states that will be predicted on. The project is very relevant since the election just happened, so it will be interesting to see what insights can be seen.

Things I liked:

1. A lot of good preprocessing ie. using logGDP rather than GDP itself
2. I liked how you split predictions into geographic regions, interesting preliminary conclusion about models trained on different regions vs. model trained on all the data
3. Clear explanations in each of the sections, well organized report & notebook

Things for improvement:

1. I notice in model-construction that there were predictions done for specific regions, it might potentially also be useful to predict using models for a state-wide basis rather than across regions, since the results for a state are determined by the counties in that state, so it could be useful to isolate predictions in that way.
2. Not sure if this would be possible but it could also be useful to factor in some data regarding red/blue party platforms, as these varied a lot in the elections.
3. Since the 2020 election occurred under very different circumstances as the previous elections (pandemic), will you be able to factor this into your model in any way?