This project delivers an end-to-end Sentiment Analysis and Business Intelligence solution for Amazon product reviews.
The solution combines: - SQL Data Cleaning - Python NLP & Machine Learning - Sentiment Classification - Interactive Power BI Dashboard
The objective is to transform unstructured customer reviews into actionable business insights that support data-driven decision-making.
Organizations receive thousands of customer reviews daily. Manually analyzing them is inefficient and time-consuming.
This solution helps businesses: - Identify customer satisfaction levels - Detect negative feedback trends - Monitor rating distribution - Improve product performance - Enable faster strategic decisions
- MySQL (SQL)
- Python (Pandas, NumPy)
- NLTK
- Scikit-learn
- Power BI
- Matplotlib
- Removed invalid ratings
- Handled NULL values
- Ensured ratings between 1--5
- Cleaned inconsistent records
Applied text preprocessing techniques: - Lowercasing - Removing punctuation - Removing numbers - Stopword removal - Tokenization - TF-IDF Vectorization
Built classification models to categorize reviews into: - Positive - Negative
Evaluation Metrics: - Accuracy - Precision - Recall - F1-Score - Confusion Matrix
The interactive dashboard provides: - Rating Distribution - Positive vs Negative Review Breakdown - Review Trends - KPI Cards - Dynamic Filters
- Majority of customers provide 4--5 star ratings.
- Negative reviews highlight product quality and delivery issues.
- Sentiment classification enables quick issue identification.
- Dashboard improves decision-making speed.
- Real-time sentiment monitoring
- Cloud deployment (AWS/Azure)
- Web-based dashboard integration
- Deep learning implementation (LSTM)
Abhishek Yewale
Data Science & Machine Learning