Sentiment Analysis on Mini IMDB Reviews

This project performs sentiment analysis on a small custom dataset of 20 movie reviews. Each review is labeled as either positive or negative. The goal is to build and evaluate a machine learning model that can classify sentiment from text data.

---

🔍 Dataset Description

• File: mini_imdb_reviews.csv
• Total Reviews: 20
• Labels: positive / negative
• Source: Custom-made for prototyping

---

🧠 Project Pipeline

1. Text Preprocessing:

   • Lowercasing, punctuation removal
   • Tokenization using regex tokenizer
   • Stopword removal
   • Lemmatization

2. Feature Extraction:

   • TF-IDF Vectorization

3. Modeling:

   • Logistic Regression

4. Evaluation:

   • Classification report (precision, recall, F1-score)
   • Confusion matrix visualization

---

🖥 How to Run

1. Install Dependencies

pip install -r requirements.txt
python sentiment_analysis.py
The output confusion matrix will be saved in visuals/confusion_matrix.png.