YEAB-G/fintech-reviews-ai-insights

AI-powered NLP pipeline for analyzing Google Play Store reviews of Ethiopian fintech banking apps. Includes automated scraping, data cleaning, sentiment analysis, theme extraction, and visual insights to support product decision-making. # 📊 Fintech Reviews AI Insights ### **AI-Powered Sentiment & Theme Analysis for Ethiopian Banking Apps** This project uses **AI, NLP, and automated data pipelines** to extract insights from thousands of Google Play Store reviews for three major Ethiopian fintech mobile banking apps: * **CBE Mobile Banking (Commercial Bank of Ethiopia)** * **BOA Mobile Banking (Bank of Abyssinia)** * **Dashen Mobile Banking (Dashen Bank)** The system performs **scraping → cleaning → sentiment analysis → theme extraction → visualization**, turning raw customer feedback into actionable insights for engineering and product teams. --- ## 🧠 What This Project Does This project solves the challenge of analyzing **unstructured customer feedback** at scale. ### ✔ Automatically scrapes Google Play reviews Fetches thousands of reviews using official app IDs and multiple locales (EN/AM). ### ✔ Cleans & preprocesses data Removes duplicates, normalizes dates, filters empty reviews, and produces a standardized dataset. ### ✔ Performs AI-driven sentiment analysis Two options: * **VADER** (fast baseline — POS / NEU / NEG) * **DistilBERT transformer** (more accurate English sentiment) ### ✔ Extracts themes using NLP Groups reviews into categories such as: * Account Access / Login issues * Performance & Speed * Reliability & Crashes * UI/UX * Feature Requests ### ✔ Generates visual insights Inside `task2_analysis.ipynb`, you get: * Rating distribution * Sentiment distribution * Top themes * Pie charts * Sample negative reviews ### ✔ Maintains professional Git workflow Branches: * `main` * `task-1-data-preprocessing` * `task-2-nlp-analysis` --- ## 📁 Project Structure > ⚠ **Important Note:** > On GitHub, some folders (like `data/`) may appear **empty**. > This is **normal**, because `.gitignore` prevents large CSV files from being uploaded. > > In VS Code locally, the folders show correctly because they contain your data. > On GitHub, only `.gitkeep` files show to preserve folde …