PRML-CSL2050 Course Project

Built With

• Scikit-learn - ML library used
• Tensorflow Keras - ML library used
• Pytorch - ML library used
• ReactJS - Javascript framework used
• Pandas - Python Data Manipulation library used
• Seaborn - Data Visualization library used

Code Structure

1. Pill_Recommendation.ipynb

This is the main file with all the preprocessing, EDA and Machine learning and Deep Learning Models.

• Installing libraries and Dependencies
• Importing dataset - UCI ML Drug Review Dataset
• Exploratory Data analysis
• Data preprocessing - Basic data information, cleaning up the data
• Dividing into test and train and transforming using Count Vectorise
• Applying Machine Learning models
• Applying Deep learning Models
• Applying Harvard Sentiment Dictionary Analysis
• Classifier Combination - Voting

2. Emotional_Analysis.ipynb

This contains the emotional analysis done on the reviews using NRC Lexicon Library.

• It contains the same preprocessing as the above file.
• Post that NRC Lexicon library is explored.
• Reviews are passed to the library functions to get the emotion scores.

How to run

1. Run the Pill_Recommendation.ipynb file first.

• The SVM code keeps crashing hence those cells should be avoided while running.
• LSTM takes about 1.5 hrs to complete running.
• The predictions from all the models are collected and stored in a .csv file.
• The final prediction scores calculated are also stored in a .csv file at the end.

2. Run the Emotional_Analysis.ipynb file after that.

• It is a completely separate entity from the Pill_Recommendation.ipynb file. The results from both the files are used to predict data based on the reviews and rating as shown on the deployed website.
• It takes 6 hours to run.

Collaborators

Name	Year	Branch
Harsh Agarwal	Sophomore	EE
Aditi Goyal	Sophomore	EE
Darshit Jain	Sophomore	EE