🔵 K-Means Clustering — University Segmentation (Unsupervised Learning)

An unsupervised machine learning project that applies K-Means clustering to group universities into categories (Private vs Public) based on their features — without using labels during training.

This project demonstrates clustering, feature scaling, and cluster evaluation techniques.

---

📌 Overview

This project uses the K-Means clustering algorithm to segment universities into two groups based on institutional features. Although the dataset contains true labels (Private/Public), the model is trained without using them, and results are later compared for evaluation.

This highlights how unsupervised learning can discover natural structure in data.

---

🧠 Learning Type

Unsupervised Learning

• No labels used during model training
• Clusters formed purely from feature similarity
• True categories used only for post-cluster evaluation

---

🎯 Objective

Group universities into clusters based on characteristics such as:

• Enrollment
• Spending per student
• Graduation rate
• Faculty ratio
• Out-of-state tuition
• Other institutional metrics

Then compare clusters with actual categories.

---

🧩 Tech Stack

• Python
• Pandas
• NumPy
• Scikit-learn
• Matplotlib
• Seaborn
• Jupyter Notebook

---

📊 Dataset

University dataset containing institutional statistics and features.

Includes:

• Numeric institutional metrics
• Private/Public label (used only for evaluation — not training)

---

🔬 Project Workflow

1️⃣ Data Exploration

• Dataset inspection
• Feature distributions
• Correlation visualization
• Pair plots and cluster tendency checks

---

2️⃣ Data Preprocessing

• Feature selection
• Feature scaling using StandardScaler
• Normalization for distance-based clustering

---

3️⃣ Model Training — K-Means

• Choose number of clusters (K = 2)
• Fit K-Means on scaled features
• Assign cluster labels --> KMeans(n_clusters=2)

---

4️⃣ Evaluation (Post-Clustering)

Since this is unsupervised learning:

• Compare cluster assignments vs actual labels
• Build confusion matrix
• Analyze cluster alignment

---

📈 Results

classification_report:

              precision    recall  f1-score   support

           0       0.21      0.65      0.31       212
           1       0.31      0.06      0.10       565

    accuracy                           0.22       777
   macro avg       0.26      0.36      0.21       777
weighted avg       0.29      0.22      0.16       777

Verdict - Not so bad considering the algorithm is purely using the features to cluster the universities into 2 distinct groups! This is how K Means is useful for clustering un-labeled data.

---

⚖️ Key Insight

Even without labels, K-Means was able to separate institutions into meaningful groups based on financial and academic features.

This demonstrates the power of unsupervised learning for structure discovery.

---

▶️ How to Run

Clone repo

git clone https://github.com/rohitb281/k-means-clustering-project.git
cd k-means-clustering-project

Install dependencies

pip install pandas numpy scikit-learn matplotlib seaborn

Run notebook

jupyter notebook

Open

K Means Clustering Project.ipynb

Run all cells.

---

🧠 Concepts Demonstrated

• Unsupervised learning
• K-Means clustering
• Distance-based algorithms
• Feature scaling importance
• Cluster evaluation
• Post-cluster label comparison

---

🚀 Possible Improvements

• Elbow method for optimal K
• Silhouette score analysis
• PCA visualization of clusters
• Try hierarchical clustering
• DBSCAN comparison
• Cluster profiling

---

⚠️ Limitations

• K must be chosen manually
• Sensitive to feature scaling
• Assumes spherical clusters
• Results depend on initialization

---

📄 License

• Open for educational and portfolio use.

---

👤 Author

• Rohit Bollapragada
• GitHub: https://github.com/rohitb281