README.md

BoxOfficePrediction

The project aims to predict the box office and score of an upcoming movie

Group Members:

Quan Yuan, Yuheng Liu, Wenlong Wu

Instruction:

data

This directory stores all the data set used and generated in the project.

1. "boxoffice_dataset.csv": the raw dataset
2. "prediction.csv": the final result predicted by the model
3. "trainset.csv": training dataset
4. "testset.csv": testing dataset

data_processing

This directory stores the codes used for data cleaning and integration.
"boxoffice_dataset.csv" in the 'data' folder has already been cleaned and integrated.

visualization

This directory stores the codes used for data analysis. They presents the relationship between some features and the movie box office.

WebScraping

This directory stores the codes used for scraping raw dataset from multiple websites. We also built a proxy program to avoid Anti-Spider.

How to run:

feature_engineering.py: This file contains all functions used for processing each column in the raw dataset.

feature_selection.py: This file generates a new dataset which is fully prepared and divides the dataset into training set and testing set. By commenting some rows, the program can generate different combinations of features.
Run this file and the generated dataset will be saved under the "data" folder.

modeling.ipynb: This notebook builds the final model and has all the material used for model tuning. Running this notebook will generate a .pkl file under the root folder.

model_testing.py: This file runs the testing using the generated model and prints the r2_socre on the console. The result is saved under the "data" folder.