Sentiment Analysis on Product Reviews

Overview

This project analyzes customer reviews from an e-commerce dataset and classifies them into Positive, Negative, and Neutral sentiments using TextBlob.

---

Objective

To automate sentiment analysis of product reviews and extract meaningful insights about customer satisfaction.

---

Dataset

• Source: Amazon Fine Food Reviews (Kaggle)
• Used: First 5000 rows
• Features:
  • Text → Review content
  • Score → Rating

---

Methodology

1. Data Cleaning

   • Removed null values and duplicates
   • Filtered relevant columns

2. Text Processing

   • Lowercasing and basic cleaning

3. Sentiment Analysis

   • Used TextBlob polarity
   • Classified into:
     • Positive
     • Negative
     • Neutral

4. Visualization

   • Bar Chart (Sentiment distribution)
   • Pie Chart (Percentage)
   • Rating vs Sentiment comparison

---

Results

• Majority of reviews are Positive (~88.34%)
• Negative reviews highlight issues like product quality and delivery delays
• Some high-rated products still show negative sentiment

---

Key Insights

• Customers are generally satisfied
• Negative feedback reveals improvement areas
• Sentiment does not always align with ratings

---

Project Structure

SentimentAnalysis_Shibom/ ├── README.md ← create this ├── analysis.ipynb ├── Reviews.csv ├── summary.pdf └── charts/

Visualizations