KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh-plugin #dsh #codex #developer-tools #ai-agents #claude-code #windows #deepseek-harness #desktop-app #typescript #rust #ai-agent

Issue-Severity-Prediction

★ 49 Open GitHub ↗

Issue Severity Prediction leverages machine learning to classify and prioritize reported issues based on severity levels. Using textual and numerical features, the project enables efficient triaging, faster resolution, and improved workflow management in software development and support systems.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# 🧠 Issue Severity Prediction using Machine Learning and BERT

This project demonstrates how to predict the severity of software issues (bugs, incidents, feature requests) using synthetic data. It includes two approaches:
- A **traditional ML approach** using TF-IDF and Random Forest.
- A **deep learning approach** using BERT from Hugging Face Transformers.

---

## 📁 Project Structure

├── issue_severity_synthetic_data.xlsx # Synthetic dataset (generated) ├── traditional_model.py # Random Forest model using TF-IDF ├── bert_model.py # Fine-tuned BERT model ├── issue_severity_model/ # Saved Hugging Face BERT model ├── issue_severity_model.pt # PyTorch format saved model └── README.md # Project documentation

---

## 🧪 Dataset

The dataset is synthetically generated and includes:
- `Title`: Short description of the issue
- `Description`: Longer explanation of the issue
- `Severity`: Labeled as one of `Low`, `Medium`, `High`, or `Critical`

To generate it, run:

```python
python generate_data.py

Or load from Excel:

import pandas as pd
df = pd.read_excel("issue_severity_synthetic_data.xlsx")

🔍 Model 1: Traditional ML

Preprocessing: TF-IDF vectorization

Model: RandomForestClassifier

Evaluation: Classification Report + Confusion Matrix

python traditional_model.py

🤖 Model 2: Deep Learning with BERT

Pretrained model: bert-base-uncased

Fine-tuned for multi-class classification

Hugging Face Trainer API for training & evaluation

python bert_model.py

Model saved to:

Hugging Face format: issue_severity_model/

PyTorch .pt format: issue_severity_model.pt

📊 Evaluation

Includes metrics:

Accuracy

F1-Score (macro)

Confusion Matrix

Classification Report

🚀 Requirements

Install dependencies:

pip install pandas scikit-learn transformers datasets torch

📦 Export

Download Excel file: issue_severity_synthetic_data.xlsx

Exported models in Hugging Face and PyTorch formats

📌 Future Improvements

Add REST API using FastAPI

Integrate with Jira/GitHub Issues for real data

Deploy model using Docker & Hugging Face Spaces
Author
  Name: Otutu Anslem
  Github: https://github.com/Otutu11
  LinkedIn: https://www.linkedin.com/in/otutu-anslem-53a687359/

📜 License

MIT License – feel free to use, modify, and distribute.

Issue-Severity-Prediction

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README