StackOne: Technology Domain Classification

A machine learning model that classifies Stack Overflow questions into technology domains based on tags and metadata.

Dataset link: https://www.kaggle.com/datasets/stackoverflow/stacklite

Model Performance

• Accuracy: 99%
• Classes: 6 technology domains
• Dataset: 48K Stack Overflow questions

Technology Domains

• Backend/Systems: Java, C#, C++, Go, Rust
• Web Development: HTML, CSS, JavaScript, React, Angular
• Data Science: Python, ML, Pandas, TensorFlow
• Database: SQL, MySQL, PostgreSQL, MongoDB
• DevOps/Cloud: Docker, AWS, Kubernetes, Linux
• Mobile Development: Android, iOS, Swift, Kotlin

Quick Start

Training

python tech_domain_classifier.py

Prediction

from predict_domain import predict_domain

domain, confidence = predict_domain(['python', 'pandas'], score=10, answer_count=2)
print(f"Domain: {domain}")  # Output: Data Science

Files

• tech_domain_classifier.py - Model training
• predict_domain.py - Prediction interface
• tech_domain_model.pkl - Trained model
• data/ - Stack Overflow dataset

Requirements

pandas
scikit-learn
numpy

Dataset Structure

• questions.csv: Question metadata (ID, score, answers, dates)
• question_tags.csv: Tag associations (ID, tag pairs)

Features

• Tag presence indicators (top 50 tags)
• Question score and answer count
• Score-to-answer ratio
• Creation year
• Tag count per question