Sentiment Analysis
Python NumPy Keras scikit-learn
Description
A sentiment analysis project on tiket.com app reviews from the Google Play Store using Python. Reviews are classified into 3 classes: Positive, Neutral, and Negative based on the star rating given by users.
The dataset was collected through scraping using the google-play-scraper library with a total of 10,000 Indonesian reviews. Three training schemes are used to compare model performance:
| Scheme | Algorithm | Feature Extraction | Data Split | Testing Accuracy |
|---|---|---|---|---|
| 1 | SVM | TF-IDF | 80/20 | 85.09% |
| 2 | Random Forest | TF-IDF | 70/30 | 87.71% |
| 3 | LSTM (Bidirectional) | Tokenizer | 80/20 | 87.64% |