Assesa44/Kenya-Ecommerce-Reviews

This project seeks to analyze Kenyan online platforms and help companies, sellers and buyers understand whether reviews reflect true product quality, or if they are false, overly biased or misleading. # 🇰🇪 Kenya E-commerce Reviews: Reviews vs Reality ### 🔍 Investigating Customer Sentiment vs Star Ratings on Jumia Kenya ## 📖 Project Overview This project analyzes customer reviews from Jumia Kenya using Natural Language Processing (NLP) to examine the relationship between numerical star ratings and written customer sentiment. The original motivation was to explore whether inflated ratings or misleading reviews could distort product visibility and seller credibility. While most reviews were verified and generally positive, the analysis revealed a more nuanced insight: ratings and written sentiment do not always align, especially for mid-range (3-star) reviews. The project demonstrates how text analytics can uncover gaps between surface-level metrics and actual customer experience, a key concern for e-commerce platforms, sellers, and consumers alike. --- ## 🎯 Objectives - Scrape real customer reviews from Jumia Kenya - Clean and preprocess unstructured text data - Perform sentiment analysis using TextBlob and VADER - Compare star ratings to text-based sentiment labels - Identify mismatches between numerical ratings and review tone - Analyze trends across product categories - Build an interactive Power BI dashboard for exploratory analysis --- ## 🛠 Tools & Technologies - *Python*: Data collection, cleaning, sentiment analysis, and EDA - *BeautifulSoup & Requests*: Web scraping - *Pandas & Matplotlib*: Data manipulation and visualization - *TextBlob & VADER*: Sentiment polarity scoring - *Power BI*: Interactive dashboard development - *Jupyter Notebook*: Analysis and documentation --- ## 🧪 Key Insights - 🤖 All reviews were from *verified purchases*, boosting credibility - 📉 Some reviews showed mismatches, products with high ratings but negative review text, and vice versa - 🎯 Most mismatches were subtle: e.g. “It’s okay” given a 5-star rating - 📊 Fashion and home appliance were the most reviewed categories --- ## Conclusion - 1-star, 2-star, 4-star …