Yosi2020/Eth-Bank-Apps-Sentiment-Analysis

Sentiment analysis of user reviews from Ethiopian banks' mobile apps (CBE, BOA, Dashen Bank). This project leverages NLP and thematic analysis to identify customer satisfaction drivers, pain points, and actionable insights to improve app performance and user retention. # 📱 Ethiopian Banks Mobile App Sentiment Analysis This repository contains a complete and structured analysis of user reviews for mobile banking applications from three prominent Ethiopian banks: **Commercial Bank of Ethiopia (CBE)**, **Bank of Abyssinia (BOA)**, and **Dashen Bank**. The project includes sentiment and thematic analysis using Natural Language Processing (NLP), data management using PostgreSQL, insightful visualizations, and actionable recommendations for enhancing user satisfaction and application performance. --- ## 📁 Repository Structure ```plaintext Eth-Bank-Apps-Sentiment-Analysis/ ├── scripts/ │ ├── review_scraper.py │ ├── review_preprocessor.py │ ├── sentiment_analysis.py │ └── thematic_analysis.py │ ├── notebooks/ │ ├── 1-data_collection.ipynb │ ├── 2-sentiment_thematic_analysis.ipynb │ ├── 3-database_storage.ipynb │ └── 4-insights_recommendations.ipynb │ ├── data/ │ ├── reviews_raw.csv │ ├── reviews_cleaned.csv │ ├── reviews_with_sentiment.csv │ └── reviews_analyzed.csv │ ├── database/ │ └── bank_reviews_dump.sql │ ├── requirements.txt ├── .gitignore └── README.md ``` --- ## 🚀 Getting Started ### 1. Clone the Repository ```bash git clone github.com cd Eth-Bank-Apps-Sentiment-Analysis ``` ### 2. Install Dependencies ```bash pip install -r requirements.txt ``` --- # 🔍 Project Workflow The project workflow explicitly follows four main analytical phases: ## 📌 Phase 1: Data Collection and Preprocessing - Scraping user reviews from Google Play Store. - Cleaning data, removing duplicates, normalizing dates, and preparing CSV files for analysis. ## 📌 Phase 2: Sentiment and Thematic Analysis - Sentiment classification explicitly using the **DistilBERT** model. - Keyword extraction using **TF-IDF** with custom-defined stop words. - Assigning thematic categories explicitly to user reviews. ## 📌 Phase 3: Database Storage - Storing processed data in **PostgreSQL**, explicitly designe …