📊 Sentiment Analysis of Alexa Reviews 📌 Overview
This project performs sentiment analysis on customer reviews of Amazon Alexa products. The dataset contains reviews, ratings, and feedback labels (positive / negative). Using Natural Language Processing (NLP) techniques and Machine Learning models, we predict whether a review expresses positive or negative sentiment.
📂 Dataset
Source: Amazon Alexa Reviews Dataset (Kaggle)
Features:
rating → numeric rating (1–5)
verified_reviews → text of the review
feedback → sentiment label (1 = positive, 0 = negative)
⚙️ Project Workflow
Data Preprocessing
Removed stopwords, punctuation, and special characters
Tokenized and vectorized text using TF-IDF
Handled class imbalance
Exploratory Data Analysis (EDA)
Distribution of positive vs. negative reviews
Word clouds for positive/negative reviews
Review length analysis
Modeling
Built machine learning pipelines with:
Logistic Regression
Naïve Bayes
Random Forest
Support Vector Machine (SVM)
Evaluated using accuracy, precision, recall, F1-score