data-mining-lab

🧠 Data Mining Lab

A practical implementation of core data mining techniques including classification, clustering, and association rule mining using Python and GUI-based tools- ORANGE and WEKA.

---

🚀 Project Overview

This project covers:

• Classification using Iris dataset
• Classification using Breast Cancer dataset
• Clustering using K-Means
• Association Rule Mining using Apriori
• ORANGE (visual ML workflows)
• WEKA (GUI-based ML tool)

---

📁 Project Structure

data-mining-lab/
│
├── notebooks/
│   ├── 01_iris_classification.ipynb
│   ├── 02_breast_cancer_classification.ipynb
│   ├── 03_clustering.ipynb
│   └── 04_association_rules.ipynb
│
├── results/
│   └── plots/
│
├── orange/
│   ├── screenshots/
│   └── notes.md
│
├── weka/
│   ├── screenshots/
│   └── notes.md
│
└── README.md

---

🌸 Iris Classification

• Models: KNN, Decision Tree
• Accuracy: 100%

📊 Output

Decision Tree Confusion Matrix

KNN Confusion Matrix

🧠 Insight

The dataset is well-separated, allowing multiple models to achieve perfect accuracy.

---

🧬 Breast Cancer Classification

• Model: Logistic Regression
• Accuracy: ~96%

📊 Output

🧠 Insight

Precision, recall, and F1-score provide better understanding than accuracy alone for real-world datasets.

---

🌀 Clustering (K-Means)

• Optimal clusters: 3
• Silhouette Score: ~0.55

📊 Output

🧠 Insight

Even without labels, natural groupings emerge from the data.

---

🔗 Association Rules (Apriori)

• Custom market basket dataset used

• Example rule:

  • Milk → Bread

🧠 Insight

Association rules reveal relationships between items and can be applied in recommendation systems.

---

🟠 ORANGE Workflow

Used ORANGE to build a visual machine learning pipeline.

📸 Screenshot

---

🔵 WEKA Experiment

Used WEKA Explorer to perform classification using J48 (Decision Tree).

📸 Screenshot

---

🛠️ Technologies Used

• Python
• Pandas, NumPy
• Scikit-learn
• Matplotlib, Seaborn
• Mlxtend
• ORANGE
• WEKA

---

📊 Key Learnings

• Different algorithms perform differently based on dataset complexity
• Visualization helps understand patterns
• Evaluation metrics are crucial
• Data mining involves both implementation and interpretation
• ORANGE provides a visual workflow-based approach, while WEKA offers a GUI-based environment for experimenting with machine learning algorithms.

---

📌 Future Improvements

• Add more real-world datasets
• Build a simple UI
• Compare additional algorithms

---

▶️ How to Run

1. Clone the repository
2. Install requirements: pip install -r requirements.txt
3. Open notebooks in Jupyter or VS Code
4. Run cells sequentially

---

✨ Conclusion

This project demonstrates practical implementation of core data mining techniques and highlights the importance of choosing appropriate models and evaluation metrics.

---

👨‍💻 Author

• Keerti Lata Choudhury