🔬 FD1 — Data Mining Interface

Interactive data mining pipeline built with Streamlit — preprocessing, unsupervised clustering, PCA visualization, and evaluation metrics.

---

🌐 Live App

https://data-mining-project-cwjq4x39pan6yahegmyaxd.streamlit.app/

📋 Overview

This app was developed as a mini-project for the FD1 (Fouille de Données 1) module at the Faculty of Computer Science, USTHB (M1 Bioinformatics, 2025–2026).

It provides a complete, interactive data mining pipeline covering:

• Volet 1 — Preprocessing: load, explore, clean, normalize, and visualize your dataset
• Volet 2 — Unsupervised Clustering: run 5 algorithms side-by-side, evaluate with silhouette score, and project results onto a 2D PCA plot

---

✨ Features

📂 Volet 1 — Preprocessing

Feature	Details
File import	CSV and Excel (.xlsx, .xls)
Overview metrics	Rows, columns, numeric cols, missing values, duplicates
Five-number summary	Min, Q1, Median, Q3, Max + Mean, Std, Mode
Missing value handling	Mean / Median / Mode imputation or row removal
Duplicate removal	One-click deduplication
Normalization	Min-Max [0,1] or Z-score (μ=0, σ=1)
Visualizations	Boxplot, Histograms, Correlation Heatmap, Scatter Matrix

🧩 Volet 2 — Clustering

Algorithm	Type	Notes
K-Means	Partitioning	Fast, centroid-based
K-Medoids	Partitioning	Manual implementation — uses real data points, robust to outliers
AGNES	Hierarchical (bottom-up)	Configurable linkage (Ward, Complete, Average, Single) + Dendrogram
DIANA	Hierarchical (top-down)	Divisive counterpart to AGNES
DBSCAN	Density-based	No k required, detects noise points, k-NN distance helper plot

Additional features: Elbow curve, Silhouette score comparison, inertia bar chart, PCA 2D projections for all algorithms simultaneously.

---

🚀 Quick Start

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/fd1-data-mining.git
cd fd1-data-mining

2. Install dependencies

pip install -r requirements.txt

3. Run the app

streamlit run app.py

The interface opens automatically at http://localhost:8501

---

☁️ Deploy on Streamlit Community Cloud (free)

1. Push this repository to GitHub (make sure app.py and requirements.txt are at the root).
2. Go to share.streamlit.io and sign in with GitHub.
3. Click "New app" → select your repo → set the main file to app.py.
4. Click Deploy — your app will be live in ~2 minutes at a public URL.

Tip: The .streamlit/config.toml file in this repo pre-configures the light theme and headless server mode for cloud deployment.

---

📁 Project Structure

fd1-data-mining/
├── app.py                  # Main Streamlit application
├── requirements.txt        # Python dependencies
├── .streamlit/
│   └── config.toml         # Streamlit theme & server config
└── README.md

---

🗂️ Recommended Datasets

Dataset	Rows	Columns	Target
Heart Disease UCI	303	14	target (0/1)
Pima Indians Diabetes	768	9	Outcome
Iris	150	5	species

---

🔧 Tech Stack

Library	Version	Purpose
Python	3.10+	Core language
Streamlit	≥ 1.32	Web UI framework
Pandas	≥ 2.0	Data manipulation
NumPy	≥ 1.24	Numerical computing
Scikit-learn	≥ 1.3	Clustering & preprocessing
Matplotlib	≥ 3.7	Charts & plots
Seaborn	≥ 0.12	Heatmaps
SciPy	≥ 1.11	Hierarchical linkage & dendrogram

---

📐 Implemented Algorithms

K-Medoids (manual implementation)

The K-Medoids algorithm was implemented from scratch without any external library, selecting real data points as cluster representatives to maximize robustness to outliers:

class KMedoids:
    def fit_predict(self, X):
        # Select initial medoids randomly
        # Assign each point to nearest medoid
        # Update medoids: pick point minimizing intra-cluster distances
        # Repeat until convergence

Normalization formulas

• Min-Max: x' = (x - x_min) / (x_max - x_min)
• Z-score: x' = (x - μ) / σ

Silhouette Score

s(i) = (b(i) - a(i)) / max(a(i), b(i)) — ranges from −1 to +1, higher is better.

---

📸 Interface Preview

The app has 3 tabs:

• 🏠 Home — feature overview and usage guide
• 📂 Preprocessing — data loading, cleaning, normalization, visualization
• 🧩 Clustering — algorithm configuration, metrics, PCA projections

---

👤 Author

KHELIF Hadil — M1 Bioinformatics, USTHB
AI & Data Science Department Academic year: 2025–2026

---

📜 License

This project is released under the MIT License.