Product Review Analysis and Visualization Using Network Graphs and Sentiment Analysis

Problem Statement

Objective: “To analyze customer reviews of various products, identifying sentiment trends, average ratings, and reviewing patterns. The analysis uses network graphs to visualize relationships between frequently reviewed products.”
Data Collection and Methodology

Data Source: The dataset, in JSON format, contains customer reviews, ratings, product identifiers (ASINs), and reviewer IDs.
Data Processing Steps:
Loaded and processed JSON data, extracting key fields like product ID, rating, review text, and reviewer ID.
Cleaned and tokenized review text for sentiment analysis and keyword clustering.
Key Analyses and Visualizations

Review Count and Average Rating by Product:

Visualization: Network graph displaying each product (node) with connections to others based on review frequency.
Insight: Visualizes relationships between products that tend to be reviewed together, giving insight into frequently paired items.
Sentiment Analysis:

Technique: TextBlob was used to analyze review sentiment, classifying each review as positive, neutral, or negative.
Result: Breakdown of positive, neutral, and negative reviews to gauge customer sentiment for each product.
Top Words in Reviews by Clustering:

Method: Used KMeans clustering to group common words in reviews, identifying key themes or concerns among customers.
Insight: Common keywords provide insights into customer expectations, product strengths, or areas of improvement.
Technical Challenges and Solutions

Graph Complexity: Ensured accurate product connections while managing large datasets with NetworkX.
Sentiment Adjustments: Adjusted sentiment analysis based on review ratings to improve accuracy.
Tools and Technologies Used

Languages & Libraries: Python, NetworkX, Matplotlib (for visualization), TextBlob (for sentiment analysis), KMeans clustering (for keyword grouping).
Platform: Google Colab for data processing and visualization.
Learning Outcomes

Gained hands-on experience with graph theory for analyzing customer patterns, refined sentiment analysis techniques, and strengthened my ability to uncover insights from unstructured text data.