The goal of this experiment is to classify if the IMDB reviews are positive or negative.
This experiment **Extract N-Gram Features** from the text and used the features to predict the sentiment of the reviews. The features are used to train multiple models : 1. Two-Class Logistic Regression 2. Two-Class Neural Network (1 Hidden Layer with 2x2500 nodes - Cost Function: tanh) 3. Two-Class Boosted Decision Tree 4. Two-Class Locally-Deep Support Vector Machine 5. Two-Class Decision Forest After running the experiment, you can see that **Two-Class Logistic Regression** has a higher accuracy and relatively lower training time than the other models. Dataset: https://www.kaggle.com/iarunava/imdb-movie-reviews-dataset