SentimentPro: Sentiment Analysis Web App

SentimentPro is a modern Flask web application for analyzing the sentiment of user reviews. Upload text or Excel/CSV files, and visualize sentiment breakdowns with beautiful interactive charts.

Features

• Instant Sentiment Prediction: Enter a review and get real-time sentiment analysis (Good, Bad, Neutral).
• Bulk Analysis via Excel/CSV Upload: Upload files and see a breakdown of review sentiments.
• Interactive Charts: Visualize sentiment distribution with clear, modern charts.
• Clean, Responsive UI: Built with Bootstrap and custom styles for a seamless experience.

Architecture

SentimentPro is a modular Flask web app with a clear separation between the web interface, preprocessing, and machine learning model. It uses NLTK for text processing and scikit-learn for sentiment prediction.

About the Model & Training Data

• Model Type: Multinomial Naive Bayes classifier (scikit-learn)
• Training Data: Drugs.com Drug Review Dataset, combining both train and test splits.
  • UCI Machine Learning Repository
  • Kaggle Dataset
• Data Size: ~215,000+ user reviews of medications.
• Labeling:
  • Reviews are labeled as:
    • positive if the rating is 8 or above
    • negative if the rating is 4 or below
    • neutral otherwise
• Text Processing:
  • Reviews are cleaned, tokenized, lowercased, and filtered for stopwords/punctuation using NLTK.
  • Features are extracted using CountVectorizer.
• Prediction:
  • The model predicts whether a review is positive, negative, or neutral based on the text content.

Why Drugs.com Data?

• It is a real-world, large-scale dataset of user experiences with medications, making it ideal for demonstrating general sentiment analysis workflows.

Note

• The model is generic and can be retrained on your own labeled review data for other domains (e.g., product reviews, movie reviews, etc.)

High-Level Architecture

flowchart TD
    User["User (Web Browser)"] -->|HTTP| FlaskApp["Flask Web App"]
    FlaskApp -->|Text Input| Preprocess["Text Preprocessing (NLTK)"]
    Preprocess -->|Features| Model["Trained ML Model (scikit-learn)"]
    Model -->|Prediction| FlaskApp
    FlaskApp -->|HTML Response| User
    FlaskApp -->|Excel Upload| Pandas["Pandas Data Processing"]
    Pandas --> Preprocess

Loading

Review Prediction Flow

flowchart LR
    A["User submits review"] --> B["Text cleaned & tokenized (NLTK)"]
    B --> C["Stopwords & rare words removed"]
    C --> D["Vectorized (CountVectorizer)"]
    D --> E["Prediction (MultinomialNB)"]
    E --> F["Result shown in UI"]

Loading

Getting Started

Prerequisites

• Python 3.8+
• pip
• (Recommended) Create and activate a virtual environment

Installation

pip install -r requirements.txt

Running Locally

# Activate your virtual environment if needed
export FLASK_APP=app.py
flask run

Visit http://127.0.0.1:5000 in your browser.

Deployment on Render

You can deploy SentimentPro to Render as a free or paid web service.

1. Ensure these files are present in your repo root:
   • app.py, requirements.txt, Procfile, render.yaml, Model.pickle, cv1, text_utils.py, and the templates/ and static/ folders.
2. Push your code to GitHub.
3. Create a new Web Service on Render:
   • Connect your GitHub repo
   • Render will auto-detect render.yaml and build your app
   • The service will use gunicorn app:app to start (see Procfile/render.yaml)
4. Access your app at the Render-provided URL.

Note: Model.pickle and cv1 must be present in the root directory for the app to work.

Production Deployment (other platforms)

• Use a WSGI server like Gunicorn (see Procfile for Heroku/Render)
• Ensure Model.pickle and cv1 are present in the root directory

Folder Structure

├── app.py                # Main Flask application
├── text_utils.py          # Text processing utilities
├── requirements.txt      # Python dependencies
├── Procfile              # For deployment (Heroku/Render)
├── Model.pickle          # Trained ML model (required for prod)
├── cv1                   # Trained vectorizer (required for prod)
├── static/
│   ├── icon.png          # App icon
│   ├── style.css         # Custom styles
│   └── ...               # Other images/assets
├── templates/
│   ├── home.html         # Main page
│   ├── pie_chart.html    # Excel chart page
│   ├── result.html       # Prediction result page
│   └── about.html        # About page

Credits

• Developed by Atul Kumar and contributors.
• Built with Flask, scikit-learn, pandas, Chart.js, Bootstrap.

---

For questions or contributions, open an issue or pull request.