Capstone Project - TrekPredict: Predicting IMDB Ratings Using NLP & Machine Learning

What if you could predict the ratings from the script of any show? I explored and tested this upon one of my most favourite tv shows in existence, Star Trek: The Next Generation.

The project contains (in viewing order):

Summary

1. KatyaKogan_Capstone_Report.pdf (final report)
2. KatyaKogan_Final_Presentation.pdf (final presentation for tech audience)
3. KatyaKogan_Demo_Day.pdf (presentation file for demo day)

Notebooks

1. Part1_TrekPredict_CleaningEDA.ipynb
2. Part2_TrekPredict_Modelling.ipynb

Data

1. model_comp.csv (comparing models)
2. PCT_graph.csv (modelling dataset)
3. TNG.csv.gz (original dataset -> https://github.com/RTrek/startrekTNGdataset)
4. total_word_count.csv

Required libraries:

• pandas
• numpy
• seaborn
• matplotlib
• sklearn

Extras:

• shap (https://shap.readthedocs.io/en/latest/index.html)
• mord (https://pythonhosted.org/mord/)
• time