Santostang / box-office-prediction

Cornell ORIE 4741 Course Project: Machine Learning with Big Messy Data
6 stars 5 forks source link

Peer Review #7

Open annasxu opened 6 years ago

annasxu commented 6 years ago

This project attempts to predict box office success using data about the movie.

I was most impressed by how this group is acquiring the data using APIs and scraping data from website. Working off a dataset that not everyone has is really exciting!

A few things I wish were different: A large majority of the Midterm Report was a list of variables. I wish I had seen less a list of variables, and more explanation on which variables they were going to choose. I saw that the group did some visualizations, but why those specifically? What else did they test? What was correlated and what was not? I also think going back to the Project Proposal, one of the requirements was an explanation of why this question is important and what larger implications does answering the question have? I think while predicting box office success is a significant question, I'm not sure if using Rotten Tomatoes ratings and other success metrics to do so is as ground-breaking. From a movie producer's perspective, wouldn't I want to predict box office success based on the appearance of certain actresses or the genre so I can make important movie decisions? Not another success metric?