Sentiment Analysis

A deployable NLP project built from NLP_NEW.ipynb, with a trained text-classification pipeline, a Streamlit front-end, and free-hosting deployment support through Streamlit Community Cloud.

Overview

This project turns a notebook-based emotion classification workflow into a small application that can be trained locally and served on the web.

It includes:

• A reproducible training script
• A saved model artifact for inference
• A Streamlit UI for single-text prediction
• Tests for training, inference, and UI helpers

Features

• Train an emotion classifier from train.txt
• Preprocess text using the notebook’s cleaning flow
• Save and reload the trained model with joblib
• Predict from a single sentence in Streamlit
• Show both predicted label and confidence score

Architecture

Training flow

train.txt -> train_model.py -> TF-IDF + Logistic Regression pipeline -> artifacts/emotion_model.joblib

Inference flow

User input -> app.py -> predict.py -> saved model artifact -> predicted label + confidence

App flow

1. User enters text in the Streamlit app
2. Input is validated
3. The saved model is loaded
4. The model predicts the most likely emotion
5. The app displays the label and confidence

Project Structure

python/
├── app.py
├── predict.py
├── train_model.py
├── train.txt
├── NLP_NEW.ipynb
├── requirements.txt
├── README.md
├── artifacts/
│   ├── .gitkeep
│   └── emotion_model.joblib
├── tests/
│   ├── conftest.py
│   ├── test_app.py
│   ├── test_predict.py
│   ├── test_project_scaffold.py
│   └── test_train_model.py
└── docs/
    └── superpowers/
        ├── plans/
        │   └── 2026-04-16-nlp-streamlit-app.md
        └── specs/
            └── 2026-04-16-nlp-streamlit-design.md

Local Setup

Install dependencies:

python3 -m pip install --user -r requirements.txt

Run the full test suite:

python3 -m pytest -v

Train the model artifact:

python3 train_model.py

Start the Streamlit app:

python3 -m streamlit run app.py

Model And App Commands

Train:

python3 train_model.py

Expected output:

Saved model to artifacts/emotion_model.joblib
Accuracy: 0.8638

Run app:

python3 -m streamlit run app.py

Deployment

GitHub

Push main to GitHub:

git push -u origin main

Streamlit Community Cloud

1. Open https://share.streamlit.io
2. Sign in with GitHub
3. Click Create app
4. Select repository sumitjadhav1703/Sentiment_analysis
5. Select branch main
6. Set main file path to app.py
7. Choose a Python version supported by Streamlit Cloud
8. Click Deploy

Notes

• The notebook NLP_NEW.ipynb is included as the original exploration and experimentation source.
• The training and deployment plan is included in docs/superpowers/plans/2026-04-16-nlp-streamlit-app.md.
• The design spec is included in docs/superpowers/specs/2026-04-16-nlp-streamlit-design.md.