Midterm report review

[mchao409]

The goal of this project is to predict whether a person will want to match with a partner given a set of attributes. The dataset is taken from a Columbia University study. In their midterm report the team has made preliminary models with Linear Regression and Logistic Regression, and plans to also explore other models moving forward.

Three things I liked:

1. I liked how you used correlation to determine whether it would be fine to aggregate the data
2. Good preliminary results already, particularly with linear regression
3. Nice explanations of how you did preprocessing with the features

Three things for improvement:

1. When you define your prediction match_rate, you only take into account the partners that would like to match with the individual. Should you also take into the account the partners that the individual would want to match with? ie. the match should be reciprocated?
2. I had a little bit of trouble understanding the graphs ie. who are the average ratings determined by in each of the graphs? It could have been helpful to include a bit of explanation.
3. It is a bit unclear to me how you will plan to combine the results, since it appears that there are a couple different predictions (ie. both the match rate and the classification)