IRIS-ML-PIPELINE

A fully modular, end-to-end Machine Learning pipeline for classifying Iris flower species using three classical ML algorithms:

• Logistic Regression
• Decision Tree Classifier
• Support Vector Classifier (SVC)

This project includes full preprocessing, training, evaluation, model comparison, and a CLI-based prediction system.
All modules are cleanly separated for reusability and easy future expansion.

---

📁 Project Structure

Iris-ML-Pipeline/
│
├── dataset/
│   ├── iris_raw.csv
│   ├── encoded.csv
│   ├── x_train.npy
│   ├── x_test.npy
│   ├── y_train.npy
│   ├── y_test.npy
│   ├── dataset_name.txt
│   └── model_comparison_iris_raw.csv
│
├── models/
│   ├── scaler.pkl
│   ├── model_LR.pkl
│   ├── model_DT.pkl
│   └── model_SVC.pkl
│
├── src/
│   ├── prepare_data.py
│   ├── train_LR.py
│   ├── train_decision_tree.py
│   ├── train_SVM.py
│   ├── predict_LR.py
│   ├── predict_decision_tree.py
│   ├── predict_SVM.py
│   └── model_compare.py
│
├── samples/
│   └── Visualisation_iris_raw.py
│
├── main.py
├── requirements.txt
└── README.md

---

🚀 Features

✔ Complete Preprocessing Pipeline

• Reads raw CSV (iris_raw.csv)
• Label-encodes species
• Train/test split (configurable)
• Feature scaling via StandardScaler
• Saves:
  • Encoded dataset
  • Scaled .npy arrays
  • scaler.pkl
  • Dataset name text file

---

✔ Three ML Models with Evaluation

Each model script computes:

• Training time
• Batch inference time
• Single-sample inference time
• Accuracy
• Confusion Matrix
• Classification Report

All trained models are saved in /models/ as:

model_LR.pkl
model_DT.pkl
model_SVC.pkl

---

✔ Model Comparison Script

model_compare.py:

• Loads all 3 trained models
• Evaluates on the same test set
• Prints structured comparison table
• Saves CSV as:

model_comparison_<dataset_name>.csv

Example:

model_comparison_iris_raw.csv

---

✔ CLI-Based Prediction System

Run main.py:

1. Logistic Regression
2. Decision Tree Classifier
3. Support Vector Classifier
4. Exit

The chosen script:

• Takes 4 numeric inputs
• Scales using scaler.pkl
• Predicts species
• Decodes class number → species name

---

✔ (Optional) Dataset Visualization

Visualisation_iris_raw.py performs simple EDA:

• Head of dataset
• Info & statistics
• Seaborn pairplot showing feature relationships

Run:

python3 samples/Visualisation_iris_raw.py

---

🧠 Models Used

1. Logistic Regression

Lightweight baseline classifier.

2. Decision Tree Classifier

Non-linear, interpretable classifier.

3. Support Vector Classifier (SVC)

Margin-based classifier suitable for smaller datasets.

---

🔧 Technologies Used

• Python 3.12
• NumPy
• Pandas
• Scikit-learn
• Matplotlib
• Seaborn
• Pickle

---

🏃‍♂️ How to Run

1. Install dependencies

pip install -r requirements.txt

2. Prepare the dataset

python3 src/prepare_data.py

This will:

• Encode dataset
• Scale features
• Save training & testing arrays
• Save scaler
• Save dataset name

3. Train all models

python3 src/train_LR.py
python3 src/train_decision_tree.py
python3 src/train_SVM.py

4. Run model comparison (optional)

python3 src/model_compare.py

5. Start CLI prediction

python3 main.py

---

🔍 Example CLI Output

🌸 IRIS FLOWER CLASSIFICATION SYSTEM 🌸
---------------------------------------

Choose a model:
1. Logistic Regression
2. Decision Tree Classifier
3. Support Vector Classifier
4. Exit

---

📌 Notes

• Re-running training scripts will overwrite previous models (if user chooses to confirm the save).
• Comparison CSV names are generated dynamically based on original dataset name.
• You can plug in any dataset with the same structure (4 features + label).
• Scripts are modular and can be reused for other ML projects.

---

📝 License

This project is made for educational and experimental purposes.
Feel free use it.