Skip to content

README.md

Latest commit

 

History

History
81 lines (47 loc) · 1.73 KB

README.md

File metadata and controls

81 lines (47 loc) · 1.73 KB
Raw

Build-a-Live-Shareable-Machine-Learning-Web-App

💬 Twitter Sentiment Analysis using Sentiment140

This project uses the Sentiment140 dataset (1.6M tweets) to train a machine learning model that predicts tweet sentiment as positive, neutral, or negative.
It also includes an interactive Streamlit web app for live predictions and batch CSV uploads.

Features

  1. Clean & preprocess raw tweets
  2. TF-IDF + Logistic Regression with hyperparameter tuning
  3. Confusion Matrix visualization
  4. Streamlit UI for:
  • Live tweet prediction
  • CSV upload for bulk sentiment analysis
  • Confidence visualization bar charts

Install Requirements

pip install -r requirements.txt

Download Dataset

Get Sentiment140 dataset from:

https://www.kaggle.com/datasets/kazanova/sentiment140

Train Model

python train_sentiment_model.py

Outputs:

sentiment_model.pkl

confusion_matrix.png

Run Web App

streamlit run app.py

Model Info

Component Description
Algorithm Logistic Regression
Vectorizer TF-IDF (10,000 features)
Evaluation 3-fold GridSearchCV
Classes Negative, Neutral, Positive

Deployment

Streamlit Cloud

Push your repo to GitHub

Go to https://share.streamlit.io

Connect to the repo and deploy

https://v2jopv7xkmsqkr4gk4cjwv.streamlit.app/

Example Output

Input:

“I absolutely love this new phone 😍🔥”

Prediction:

✅ Sentiment: Positive Confidence: 97%

image

Author

Ittyavira C Abraham

MCA (AI) — Amrita Vishwa Vidyapeetham