A Machine Learning project to predict star ratings from Yelp reviews. This program uses five different models to conduct its multiclass classifications:

• Logistic Regression
• Perceptron
• Support Vector Machine
• Nearest Centroid
• Voting Ensemble of the above four

• Report.pdf - a report explaining the project
• AgabinSeghbossianShrestha_predictions.csv - the final predictions for the provided test
• AgabinSeghbossianShrestha_code.zip - the zip file of all source code
  • pickled_feature_vectorizer - the TFID vectorizer used to extract a set of features from text instances
  • pickled_svm - pickled SVM model
  • pickled_perceptron - pickled perceptron model
  • pickled_nc - pickled nearest centroid model
  • pickled_logistic_regression - pickled logistic regression model
  • feature_extractions_BOW.ipynb - python notebook with code related to feature extraction
  • perceptron_nearest_centroid.ipynb - python notebook with code related to perceptron and nearest centroid classifiers
  • SVM.ipynb - python notebook with code related to SVM classifier
  • logistic_regression.ipynb - python notebook with code related to logistic regression model
  • predictor_client.py - python file to be used for testing models

1. Unzip AgabinSeghbossianShrestha_code.zip
2. Run predictor_client.py with the filepath to the testing dataset as the only argument to the program
3. The program will use a voting ensemble of four different models to output its predictions to predictions_group2.csv NOTE: Do not run any of the ipynb files. Many of them have long runtimes and rely on certain data files to be inside the working directory.