IMDB Movie Sentiment Analyzer

A lightweight web application that predicts the sentiment (Positive/Negative) of movie reviews. This project features a FastAPI backend and a custom NumPy-based inference engine, removing the need for heavy machine learning libraries in the production environment.

📌 Project Overview

This application takes a movie review as input and outputs the probability of it being a positive review. It uses a neural network trained on the IMDB dataset.

📊 Dataset Details: IMDB Movie Reviews

The model was trained on the Large Movie Review Dataset, a benchmark for binary sentiment classification.

• Total Samples: 50,000 reviews (25,000 training / 25,000 testing).
• Balance: Perfectly balanced with 50% positive and 50% negative reviews.
• Feature Engineering:
  • Limited to the top 10,000 most frequent words.
  • Multi-Hot Encoding: Text is converted into a 10,000-dimensional binary vector where 1 represents the presence of a word and 0 its absence.

📈 Model Performance & Evaluation

The model was evaluated using a test subset of 200 samples. Below are the detailed metrics and the confusion matrix.

Classification Report

Class	Precision	Recall	F1-Score	Support
0 (Negative)	0.89	0.69	0.78	163
1 (Positive)	0.32	0.62	0.42	37
Accuracy			0.68	200
Macro Avg	0.60	0.66	0.60	200
Weighted Avg	0.78	0.68	0.71	200

Confusion Matrix

The following matrix visualizes the performance of the model, showing the distribution of True Positives, True Negatives, False Positives, and False Negatives.

🧠 Model Architecture

The model is a Feed-Forward Neural Network with the following structure:

1. Input Layer: Vectorized text (Bag-of-Words style, capped at 10,000 words).
2. Hidden Layers: Two hidden layers with ReLU activation.
3. Output Layer: Single unit with Sigmoid activation to output a probability between 0 and 1.

Note: The actual training happens in imdb_nn_model.ipynb, while app.py performs the inference using the saved weights.

📂 Repository Structure

.
├── app.py                  # FastAPI Backend & Inference Engine
├── imdb_architecture.png   # Architecture Diagram
├── imdb_nn_model.ipynb     # Training Notebook
├── index.html              # Main Frontend UI
├── index3.html             # Alternative Frontend
├── model_params.npz        # Trained Weights (W1, b1, W2, b2, W3, b3)
├── requirements.txt        # Dependencies
├── results/
│   └── cm.png              # Confusion Matrix Plot
└── word_index.json         # Word-to-Index Vocabulary