Midterm Report Peer Review - yj76

[Eve208]

This project explores the potential features that make a movie successful and then try to do predictions on the features and develop a model to predict the IMBD score of movies.

Things I like:

1. The project talks about some techniques that can be used to avoid over/underfitting
2. The group started preliminary analysis by fitting a linear model and gained some insights from it.
3. The report includes several visualizations of some features and there’s a clear outline for future work.

Things that I think can be improved and some concerns:

1. It seems you need to do more data cleaning before trying to fit models. You only talked about removing irrelevant features and some possible ways to deal with missing data. Also, are there any outliers in the dataset that may affect your model?
2. The report doesn’t talk about any feature engineering which I think is something you can try and it may be helpful for choosing what variables to fit.
3. You mentioned about the possibility of incorporating facebook likes/twitter favorites/Youtube views, which I think will result in a lot of missing data. And it won’t be a good idea to just simply discard the entries with the missing values since some movies with very high scores are documentaries which are very likely to be less present on social media.