Iris Classification in Python

Academic machine learning project for classifying Iris flower species using Python and scikit-learn.

Project Overview

This project applies supervised classification methods to the classic Iris dataset. The notebook includes data loading, preprocessing, train/test splitting, model training, evaluation metrics, and classification reports.

Dataset

The repository includes the Iris dataset files:

• data/iris.data
• data/iris.names
• data/bezdekIris.data

Technologies

Category	Tools
Language	Python
Notebook	Jupyter / Google Colab
Data Analysis	pandas, NumPy
Machine Learning	scikit-learn
Visualization	Matplotlib

Models and Evaluation

The notebook demonstrates classification with three algorithms, compared on the same train/test split:

• Logistic Regression
• K-Nearest Neighbors, with the optimal k selected via GridSearchCV (5-fold cross-validation)
• Support Vector Machine (RBF kernel)

Evaluation includes accuracy, a classification report, and a confusion matrix for the best-performing model.

Results

On a 70/30 stratified train/test split, KNN (tuned) reached the best accuracy:

Model	Test Accuracy
Logistic Regression	91.1%
KNN (tuned)	95.6%
SVM (RBF)	93.3%

The confusion matrix below shows the few versicolor/virginica confusions typical of this dataset — the two species overlap slightly in petal measurements:

Visualizing the decision boundary on the two most discriminative features (petal length & width) makes the class separation intuitive:

Repository Structure

iris-classification-python/
├── README.md
├── LICENSE
├── requirements.txt
├── .gitignore
├── data/
│   ├── iris.data
│   ├── iris.names
│   └── bezdekIris.data
├── assets/
│   ├── accuracy_comparison.png
│   ├── confusion_matrix.png
│   └── decision_boundary.png
└── notebooks/
    └── iris_classification.ipynb

How to Run

pip install -r requirements.txt
jupyter notebook notebooks/iris_classification.ipynb

If running from the notebook folder, adjust the dataset path to ../data/iris.data if necessary.

Skills Demonstrated

• Supervised classification
• Data preprocessing and label encoding
• Model training with scikit-learn
• Classification metrics and model comparison

Author

Manassé Makuikila Lusaku

License

MIT License