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.
After Cross Validation
- 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.
- Need better cross validation approach eg using Random SearchCV.
- Can use Voting Classifier for multiple models
- Apply Stacking for combination of algorithms.
- Incorporate additinal features in the model.