BRImo Sentiment Analysis

This project scrapes Google Play Store reviews for the BRImo mobile application and builds sentiment-analysis models to categorize user feedback. Two Jupyter notebooks cover the end-to-end workflow: large-scale data collection and multiple machine-learning experiments on Indonesian-language reviews.

Installation

1. Ensure Python 3.10+ is available.
2. Create and activate a virtual environment (recommended).
3. Install dependencies:

   pip install -r requirements.txt

Usage

• Scrape fresh reviews
  Open Scrapping_Review_BRImo_Playstore.ipynb in Jupyter, update the desired app_id or output filename if needed, and run the notebook to download reviews into a CSV file (default: ulasan_aplikasi_brimo.csv).

• Run sentiment analysis
  Launch Sentiment_Analysis_Aplikasi_BRImo.ipynb to:

  1. Load the dataset (either the provided CSV or the one generated above).
  2. Execute preprocessing that normalizes text, maps Indonesian slang, and removes noise.
  3. Label reviews using a lexicon-based approach.
  4. Explore data through plots and word clouds.
  5. Train and evaluate several classifiers (logistic regression, SVM, TF-IDF neural net, and Bag-of-Words neural net).
  6. Perform inference on custom text inputs with the best-performing model.

Project Structure

.
├── Scrapping_Review_BRImo_Playstore.ipynb   # Google Play review scraper
├── Sentiment_Analysis_Aplikasi_BRImo.ipynb  # Preprocessing, modeling, and evaluation
├── requirements.txt                         # Python dependencies
└── ulasan_aplikasi_brimo.csv                # Sample dataset of BRImo reviews

License

No explicit license is provided. Add an appropriate LICENSE file before redistributing or using this code in production.

Contributing

1. Fork the repository and create a feature branch.
2. Make changes with clear commits and, when possible, add tests or notebook annotations explaining new behavior.
3. Ensure notebooks run top-to-bottom without errors.
4. Open a pull request describing the motivation, approach, and validation steps. Feedback and review improvements are welcome.