Amazon-sentiment-analysis

This repository implements sentiment analysis on Amazon product reviews using VADER (Valence Aware Dictionary and sEntiment Reasoner) and RoBERTa (A Robustly Optimized BERT Pretraining Approach). The project classifies reviews into positive, negative, or neutral categories, offering valuable insights into customer feedback and product performance.

Sentiment Analysis on Amazon Reviews

Project Overview

This project aims to analyze customer feedback on Amazon products using sentiment analysis techniques. By extracting insights from product reviews, businesses can better understand customer sentiments and improve their offerings. The project employs both simple rule-based and advanced machine learning models to classify sentiments as positive, negative, or neutral.

Table of Contents

• Technologies Used
• Implementation Steps
• Key Findings
• How to Run the Project

Technologies Used

• Python: Primary programming language for data manipulation and analysis.
• Pandas: Library for data loading and preprocessing.
• Matplotlib & Seaborn: Libraries for data visualization.
• Natural Language Processing (NLP): Techniques for text data processing.
• VADER: A rule-based sentiment analysis tool for classifying sentiments.
• RoBERTa: A pretrained model from Hugging Face Transformers for advanced sentiment classification.

Implementation Steps

1. Data Loading & Preprocessing: Load the Amazon reviews dataset using Pandas, cleaning the data by removing missing values and irrelevant columns.
2. Exploratory Data Analysis: Visualize the distribution of review scores and sentiment trends using Matplotlib and Seaborn.
3. Natural Language Processing (NLP): Preprocess text data through tokenization, stopword removal, and text vectorization.
4. Sentiment Analysis:
   • Apply VADER sentiment analysis for a straightforward, rule-based classification.
   • Utilize the RoBERTa pretrained model for a more nuanced sentiment classification.
5. Model Evaluation: Evaluate the performance of sentiment analysis models using accuracy and F1-score metrics.

Key Findings

The analysis revealed key insights into customer sentiment, helping to identify product strengths and areas for improvement. The project demonstrated that sentiment analysis can effectively inform business strategies and enhance customer experience.

How to Run the Project

To run this project locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/BalajiV21/Amazon-sentiment-analysis.git