https://github.com/18mahi/tweet-sentiment-analysis

Classify tweets into happy, sad, angry, excited, and neutral with this interactive Python model. Combining TF-IDF text features with engineered numeric features like emoji sentiment, polarity, subjectivity, and punctuation counts, it demonstrates intermediate-level NLP, feature engineering, and visualization skills.
https://github.com/18mahi/tweet-sentiment-analysis

data-science data-visualization jupyter-notebook machine-learning matplotlib nlp numpy pandas sklearn string textblob

Last synced: 4 months ago
JSON representation

Classify tweets into happy, sad, angry, excited, and neutral with this interactive Python model. Combining TF-IDF text features with engineered numeric features like emoji sentiment, polarity, subjectivity, and punctuation counts, it demonstrates intermediate-level NLP, feature engineering, and visualization skills.

• Host: GitHub
• URL: https://github.com/18mahi/tweet-sentiment-analysis
• Owner: 18mahi
• License: mit
• Created: 2025-10-04T15:13:39.000Z (4 months ago)
• Default Branch: main
• Last Pushed: 2025-10-04T16:15:37.000Z (4 months ago)
• Last Synced: 2025-10-04T17:26:56.567Z (4 months ago)
• Topics: data-science, data-visualization, jupyter-notebook, machine-learning, matplotlib, nlp, numpy, pandas, sklearn, string, textblob
• Language: Jupyter Notebook
• Homepage:
• Size: 319 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# tweet-sentiment-analysis

This repository contains an intermediate-level tweet sentiment analysis model built using Python, Scikit-learn, TextBlob, and Plotly. The project demonstrates feature engineering, machine learning, and interactive visualization in a complete workflow, making it ideal for portfolios and learning purposes.

# Project Overview

The model classifies tweets into five sentiment categories: happy, angry, sad, excited, and neutral. It combines TF-IDF text features with intermediate-level engineered features, including:

Emoji sentiment scoring

Polarity & subjectivity (TextBlob)

Exclamation and question counts

Word count and average word length

Hashtags, mentions, and capital letter ratio

The project leverages a RandomForest classifier trained on a synthetic dataset, showcasing feature importance, prediction probabilities, and interactive visualizations for both evaluation and user inputs.



# Key Features

Text Preprocessing: Cleans tweets by removing URLs, mentions, hashtags, and punctuation.

Feature Engineering: Combines text-based TF-IDF vectors with numeric sentiment and structural features.

Model Evaluation: Includes accuracy metrics, classification report, and confusion matrix.

# Visualization:

Probability bar charts for individual predictions using Plotly.

Top feature importance visualized with Matplotlib.

Dynamic User Input: Users can type custom tweets and receive predicted sentiment and probability distribution interactively.

# Technologies & Libraries

-Python 3.x

-Pandas & Numpy

-Scikit-learn

-TextBlob

-Matplotlib & Seaborn

-Plotly

# How to Use

Run the notebook in Jupyter.

Explore the pre-generated test predictions and visualizations.

Use the dynamic input section to classify any custom tweet.

Observe probability bars and feature contributions for interpretability.

# Learning Outcomes

This project demonstrates practical NLP techniques, intermediate-level feature engineering, model evaluation, and interactive visualization, making it an excellent addition to a data science portfolio. It also provides a foundation to explore more advanced topics like bigram/trigram TF-IDF, deep learning, or sentiment analysis on larger datasets.