Peer review by Hao

[rh443453518]

Hao Rong, hr335

Three things I like
1.The core motivation of this proposal is very practical and useful. If I were the manager of a film company, a film director, or an investor, I would definitely want to look into this research and see what make a good film.
2.The project is very likely to harvest good result since the dataset is relatively rich. The fields of the dataset are substantial enough to build a good model.
3.This is an exploratory research combining entertainment industry with big data. The result might be very interesting or even unexpected.

Three Areas to improve

1. One thing I notice that you guys have two main questions which are both certainly valuable. I just want to say some movie make money even they have a low rating. Dive into that aspect might be one of another interesting direction.
2. Doing some feature engineering using human sense might be useful. For example, we can build a feature called effort that put into making the movie using the cost of the movie, the time spent on making the movie, etc. Making this kind of feature that easy to interpret might give you a better chance to measure and improve your model performance.
3. Dividing data into subgroup to analyze by genre, or by year might be very handy. For example, a certain genre of the movie will be more likely to yield profit while another genre is more likely to earn a reputation for the director.