Box-Office-Prediction-using-Movie-Trailers

This repository contains the codes and research report for the Box Office Return Prediction Research Project developed during the International Masters in Data Analytics at the University of Hildesheim.

preprocessing.ipynb

1. Preprocessing, augmentation and conveting the videos into frames

LTC_Implementation.ipynb

1. Implementation of Long Temporal 3D Convolution on movie trailers
2. Integration of LTC with FC networks on meta-data
3. Comparison with baseline linear regression and gradient boosting methods

Lstm_Implementation.ipynb

1. Implementation of Bi-directional LSTM module to train on the movie trailers
2. Integration of LSTM module with FC netowrks trained on meta-data
3. Comparison with baseline linear regression and gradient boosting methods