🌸 Iris Flower Classification & Interactive EDA

An end-to-end Machine Learning pipeline that explores, visualizes, and classifies the classic Iris Dataset. This project trains a machine learning classifier to accurately distinguish between three distinct species of Iris flowers using structural floral features.

📌 Project Architecture

This repository contains a structured analytical pipeline divided into three major stages:

1. Exploratory Data Analysis (EDA)

Understanding data relationships before modeling via high-quality data visualizations:

• Class Balance Verification: Assessing target variable distributions via frequency count plots.
• Feature Clustering Assessment: Generating a complete pair-wise scatter plot matrix (pairplot) to see structural boundaries between classes.
• Multicollinearity Analysis: Creating a custom correlation heatmap (coolwarm colormap) to identify highly correlated independent features.

2. Preprocessing & Data Engineering

• Dimensionality Reduction: Dropping the arbitrary database primary key (Id) to prevent the model from capturing a synthetic linear pattern.
• Target Label Encoding: Transforming categorical species labels (Iris-setosa, Iris-versicolor, Iris-virginica) into numerical target integers ($0, 1, 2$) using LabelEncoder.
• Stratified Splitting: Implementing a 20% test split explicitly leveraging stratify=y to preserve exact target class distributions across training sets and prevent split bias.

3. Model Implementation & Diagnostic Evaluation

• Supervised Learning: Fitting a DecisionTreeClassifier configured with reproducible global random states.
• Performance Measurement: Evaluating classification boundaries using overall accuracy, comprehensive classification summaries (precision, recall, f1-score), and visual Confusion Matrices.
• Model Interpretability: Computing exact mathematical feature importance vectors to display which geometric properties dictate flower identification.

---

🛠️ Technical Stack

Category	Tools / Libraries Used
Core Language	Python 3.x
Data Manipulation	pandas, numpy
Data Visualization	seaborn, matplotlib
Machine Learning	scikit-learn

---

📈 Pipeline Highlights & Metrics Summary

The code includes critical data science practices that protect against common model failures:

Why Stratification Matters: The use of stratify=y ensures that each flower class is equally represented in both your training set and testing set. Without this, a random split could accidentally put all of one species into the test set, completely crippling the training process.

The evaluation suite outputs the following diagnostic assets:

• Accuracy Score: Direct percentage metric of overall correct predictions.
• Classification Report: Precision and recall per class to confirm there are no weak spots in individual species identification.
• Feature Importance Chart: Visually ranks features so you can verify if biological assumptions match mathematical model weights (e.g., assessing the predictive power of Petal Width vs. Sepal Length).