Sentiment-analysis Natural Language Processing

The goal of the project is to develop a machine learning model that can accurately classify the sentiment of product reviews as either positive or negative.

Results

After Cross Validation

Findings

1. We can see that the precision is 80 % and the accuracy is around 76 % which is not desirable for the production. AUC score is around 86.2% which great for the model performance.
2. Need better cross validation approach eg using Random SearchCV.

Ways to improve model accuracy

1. Can use Voting Classifier for multiple models
2. Apply Stacking for combination of algorithms.
3. Incorporate additinal features in the model.