Medicine Review with Sentiment Analysis - ML

Category: Artificial Intelligence

Sentiment Analysis on medicine reviews:

- Data pre-processing (feature extraction, removing stop words, punctuations, stemming, lemmatization etc.).
- Run models.
- Conclude and visualize data (accuracy, confusion matrix etc.).

Binary Class:

Accuracy:
- Naive Bayes : 53 %.
- Adaboost : 89 %.
- Logistic Regression : 80 %
- CatBoost : 90.7 %.
-----------------------

Multi-class (5-class):

Class Count:
- 0 Highly Effective 1741.
- 1 Considerably Effective 1238.
- 2 Moderately Effective 572.
- 3 Ineffective 329
- 4 Marginally Effective 263.

Accuracy:
- Naive Bayes : 34 %.
- Adaboost : 44 %.
- Logistic Regression : 48 %.