Iris Flower Classification using Machine Learning

A comprehensive machine learning project for classifying Iris flower species using multiple classifiers with rigorous evaluation methodology.

---

Table of Contents

• Project Overview
• Dataset
• Methodology
• Results
• Project Structure
• Installation
• Usage
• Key Visualizations
• Future Improvements
• Author

---

Project Overview

This project implements a complete machine learning pipeline to classify Iris flowers into three species based on their morphological measurements. The workflow includes:

1. Data Cleaning — Handling missing values, duplicates, and feature selection
2. Exploratory Data Analysis — Correlation analysis, distribution visualization, and class separability assessment
3. Feature Engineering — Label encoding, train-test splitting with stratification, and feature scaling
4. Model Training — Five different classifiers (SVM, Logistic Regression, Decision Tree, Random Forest, KNN)
5. Robust Evaluation — Stratified 5-fold cross-validation with multi-metric assessment
6. Hyperparameter Tuning — GridSearchCV optimization for top-performing models
7. Model Comparison — Side-by-side performance analysis with visualizations

---

Dataset

The Iris dataset is one of the most well-known datasets in machine learning, introduced by British statistician Ronald Fisher in 1936. It contains 150 samples from three species of Iris flowers, with 50 samples per species.

Feature	Description	Unit
Sepal Length	Length of the sepal	cm
Sepal Width	Width of the sepal	cm
Petal Length	Length of the petal	cm
Petal Width	Width of the petal	cm
Species	Target class (Iris-setosa, Iris-versicolor, Iris-virginica)	—

Source: UCI Machine Learning Repository

---

Methodology

Preprocessing

• Dropped the Id column (non-informative)
• Removed duplicate rows
• Encoded species labels using LabelEncoder
• Split data 80/20 with stratification to preserve class balance
• Scaled features using StandardScaler (critical for SVM, Logistic Regression, and KNN)

Models Trained

Model	Why Included
Support Vector Machine (SVM)	Excellent for high-dimensional spaces; uses kernel trick for non-linear boundaries
Logistic Regression	Strong baseline linear classifier; interpretable coefficients
Decision Tree	Non-parametric; handles non-linear relationships; interpretable
Random Forest	Ensemble of trees; reduces overfitting; robust to noise
K-Nearest Neighbors	Instance-based learning; simple yet effective for small datasets

Evaluation Strategy

• Train-Test Split: 80/20 with stratified sampling
• Cross-Validation: 5-fold stratified CV for robust performance estimation
• Metrics: Accuracy, Precision, Recall, F1-Score, Confusion Matrix
• Hyperparameter Tuning: GridSearchCV with exhaustive parameter search

---

Results

Cross-Validation Performance

Model	Mean CV Accuracy	Std Deviation
Support Vector Machine	~97-100%	Low
Logistic Regression	~96-100%	Low
Random Forest	~95-100%	Low
Decision Tree	~95-100%	Moderate
K-Nearest Neighbors	~95-100%	Low

Note: Exact values depend on random seed and data splits. Run the notebook for precise results.

Key Findings

1. Petal dimensions are the most discriminative features — Petal length and petal width are highly correlated (r=0.96) and provide clear class separation
2. Iris-setosa is linearly separable from the other two species
3. Feature scaling is essential for distance-based algorithms (SVM, KNN)
4. SVM with RBF kernel achieves the best performance after hyperparameter tuning

---

Project Structure

Iris_Flower_Classification_MachineLearning/
│
├── Iris_Flower_Classification.ipynb   # Main Jupyter notebook
├── Iris.csv                           # Dataset
└── README.md                          # This file

---

Installation

Prerequisites

• Python 3.8 or higher
• pip or conda package manager

Setup

1. Clone the repository:

   git clone https://github.com/yourusername/Iris_Flower_Classification_MachineLearning.git
   cd Iris_Flower_Classification_MachineLearning

2. Create a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install pandas numpy matplotlib seaborn scikit-learn jupyter

4. Launch Jupyter Notebook:

   jupyter notebook

5. Open Iris_Flower_Classification.ipynb and run all cells.

---

Usage

The notebook is designed to be run sequentially from top to bottom. Each section is clearly marked with markdown headers:

1. Section 1-3: Setup and data preparation
2. Section 4: EDA with interactive visualizations
3. Section 5: Preprocessing pipeline
4. Section 6-7: Model training and cross-validation
5. Section 8: Hyperparameter optimization
6. Section 9-10: Detailed evaluation and comparison
7. Section 11: Summary and conclusions

---

Key Visualizations

The notebook generates several publication-quality visualizations:

• Correlation Heatmap — Feature relationships and multicollinearity
• Violin Plots — Distribution shape and class separability
• Pair Plot — All pairwise feature relationships colored by species
• Box Plots — Feature ranges and outlier detection
• Histograms — Feature distributions per species
• Confusion Matrices — Per-class classification accuracy
• Model Comparison Charts — Bar and box plots of cross-validation scores

---

Author

Naveed Ahmad

If you found this project useful, please consider giving it a star on GitHub!

---

License

This project is open source and available under the MIT License.