symo101/kenya-news-sentiment-tracker

NLP sentiment analysis of Kenyan news headlines scraped from StandardMedia.co.ke TextBlob scoring, category breakdown and live headline scorer deployed on Streamlit # 📰 Kenya News Sentiment Tracker > NLP project — real Kenyan news headlines scraped from StandardMedia.co.ke, analysed for sentiment using TextBlob and deployed as a live interactive dashboard. 🔗 **Live App:** kenya-news-sentiment-tracker.streamlit.app --- ## 📸 Screenshots --- ## 🎯 Project Overview This project scrapes real headlines from **StandardMedia.co.ke** across 6 news categories, scores each headline's sentiment using **TextBlob NLP**, and visualises the results in an interactive Streamlit dashboard. **Key Question:** *Is Kenyan news coverage mostly positive, negative or neutral — and which categories drive the most negative sentiment?* --- ## 🔄 Project Pipeline ``` SCRAPE ──► CLEAN ──► SENTIMENT ──► VISUALISE ──► DEPLOY ``` --- ## 📁 Project Structure ``` kenya-news-sentiment-tracker/ │ ├── app.py # Streamlit dashboard ├── standard_scraper.py # Web scraper ├── sentiment_analysis.ipynb # Cleaning + sentiment notebook │ ├── news_raw.csv # Raw scraped headlines (214 rows) ├── news_clean.csv # Cleaned + scored dataset (213 rows) │ ├── requirements.txt # Python dependencies └── README.md ``` --- ## 🔍 Stage Breakdown ### 1️⃣ Web Scraping — `standard_scraper.py` - Built with `requests` and `BeautifulSoup` - Used Chrome DevTools to identify stable article link selectors - Scraped 6 categories: Politics, Business, Health, Sports, World, Counties - Fields collected: headline text, category, URL, scraped timestamp - Output: `news_raw.csv` — **214 headlines** ### 2️⃣ Cleaning — `sentiment_analysis.ipynb` - Removed extra whitespace and special characters - Dropped duplicate and very short headlines - Parsed timestamps into proper datetime format - Output: `news_clean.csv` — **213 headlines, 0 missing values** ### 3️⃣ Sentiment Analysis — `sentiment_analysis.ipynb` - Scored every headline using **TextBlob** polarity - Polarity range: –1.0 (very negative) → 0.0 (neutra …