NLP Emotion Classification

Classifies text into one of six emotions using TF-IDF + Logistic Regression, served through a Streamlit app for real-time prediction.

😔 Sadness (0)	😠 Anger (1)	❤️ Love (2)	😮 Surprise (3)	😨 Fear (4)	😊 Joy (5)

Model Performance

Dataset	Split	Features	Classifier	Classes	Test Accuracy
16,000 samples	80 / 20 (random_state=42)	TF-IDF	Logistic Regression	6	86.16%

Accuracy is measured on a held-out test set not used during training.

Pipeline

Raw Text → Lowercase → Remove Punctuation → Remove Numbers → Remove Non-ASCII → Remove Stopwords → TF-IDF → Logistic Regression → Emotion

The same preprocessing and trained vectorizer are reused at inference time.

Tech Stack

Language	Data	ML	Serialization	App	Dev
Python	pandas	scikit-learn	joblib	Streamlit	Jupyter, VS Code

Project Structure

File	Description
app.py	Streamlit app for emotion prediction
emotion_model.pkl	Trained model, TF-IDF vectorizer and config
nlp-emotion-classification.ipynb	Model development and evaluation
train.txt	Labelled text dataset
requirements.txt	Python dependencies

Run Locally

git clone https://github.com/arshiafreen090/NLP-Emotion-Classification.git
cd NLP-Emotion-Classification
pip install -r requirements.txt
streamlit run app.py

App Features

Text input • Predicted emotion • Confidence score • Probability distribution across all classes • Processed text view • Model info

Limitations

Built for NLP learning and demonstration. Emotion is subjective, and the model may miss context or nuance. Confidence scores are not a measure of a person's actual emotional state.

---

Built with Python, scikit-learn and Streamlit.