---

⚡ Quick Start

Install MLPilot:

pip install mlpilotx

Run your first ML pipeline:

mlpilot run --data examples/USA_Housing.csv

Specify the target column manually:

mlpilot run --data examples/USA_Housing.csv --target price

---

🔥 What MLPilot Does

Instead of writing hundreds of lines of boilerplate code, MLPilot automatically performs:

Stage	Description
📂 Load	CSV loading & validation
📊 EDA	Statistics, missing values & correlations
🧹 Clean	Duplicates, sparse columns & outlier handling
⚙️ Feature Engineering	Encoding, scaling & transformations
✂️ Split	Train / Test split
🔧 Preprocess	Scikit-learn preprocessing pipeline
🏆 Compare	14+ Machine Learning models
🎯 Tune	Bayesian optimization with Optuna
🔍 Explain	SHAP feature importance & visualizations
📦 Export	Deployment-ready .joblib pipeline

---

🔄 Pipeline Workflow

                 Raw Dataset
                      │
                      ▼
              📂 Load Dataset
                      │
                      ▼
            📊 Statistical EDA
                      │
                      ▼
              🧹 Data Cleaning
                      │
                      ▼
         ⚙️ Feature Engineering
                      │
                      ▼
           ✂️ Train / Test Split
                      │
                      ▼
        🔧 Preprocessing Pipeline
                      │
                      ▼
        🏆 Compare 14+ ML Models
                      │
                      ▼
     🎯 Optuna Hyperparameter Tuning
                      │
                      ▼
      🔍 SHAP Explainability Report
                      │
                      ▼
          📈 Model Evaluation
                      │
                      ▼
   📦 Export Production Pipeline (.joblib)

---

✨ Features

• 📂 Automatic CSV support
• 🤖 Automatic Regression & Classification detection
• 📊 Smart Exploratory Data Analysis
• 🧹 Missing value & outlier handling
• ⚙️ Feature engineering pipeline
• 🏆 Cross-validation model leaderboard
• 🎯 Bayesian hyperparameter optimization
• 🔍 SHAP explainability visualizations
• 📦 Export complete inference pipeline
• 💻 Rich CLI & Python API support

---

💻 CLI Usage

Automatic Training

mlpilot run --data examples/heart.csv

Regression

mlpilot run --data examples/USA_Housing.csv --target price

Classification

mlpilot run --data examples/patient_adherence_dataset.csv --target adherence

Custom Output Folder

mlpilot run --data data.csv --output outputs/

---

🐍 Python API

from ml_pilot import PipelineRunner
from ml_pilot.config import load_config

config = load_config()

runner = PipelineRunner(config)

context = runner.run(
    data_path="examples/USA_Housing.csv",
    target="price"
)

print(context.best_model_name)
print(context.metrics)

---

📁 Example Datasets

MLPilot includes ready-to-use datasets inside the examples/ folder.

Dataset	Task
USA_Housing.csv	Regression
insurance.csv	Regression
heart.csv	Classification
patient_adherence_dataset.csv	Classification
Student_performance_data.csv	Classification
Food_Delivery_Times.csv	Regression
Exam_Score_Prediction.csv	Regression
taxi_trip_pricing.csv	Regression
personality_synthetic_dataset.csv	Classification

Example:

mlpilot run --data examples/insurance.csv --target charges

---

📦 Generated Artifacts

Every successful run generates:

mlpilot_artifacts/
├── mlpilot_pipeline.joblib
├── leaderboard.csv
├── metrics.json
├── model_comparison.json
├── feature_importances.json
├── eda_report.json
├── run_metadata.json
├── serving_schema.json
├── predict_snippet.py
├── DEPLOY.md
└── shap/
    ├── shap_summary.png
    ├── shap_dependence_*.png
    └── shap_waterfall.png

---

🧠 Supported Models

Category	Models
Linear	Linear Regression, Ridge, Lasso, ElasticNet
Tree	Decision Tree, Random Forest, Extra Trees
Boosting	Gradient Boosting, HistGradientBoosting
Instance	KNN, SVR
Neural	MLP
Classification	Logistic Regression, SGD, Linear SVC, Passive Aggressive

---

🧪 Edge Case Testing

MLPilot includes dedicated validation datasets.

tests/
└── edge_cases/
    ├── empty.csv
    ├── one_row.csv
    ├── all_null.csv
    ├── duplicate_col.csv
    ├── target_missing.csv
    ├── only_numeric.csv
    └── only_categorical.csv

Run an edge-case test:

mlpilot run --data tests/edge_cases/empty.csv

---

📂 Project Structure

MLPilot/
├── configs/
│   └── default.yaml
├── examples/
│   ├── USA_Housing.csv
│   ├── insurance.csv
│   ├── heart.csv
│   └── ...
├── src/
│   └── ml_pilot/
│       ├── cli.py
│       ├── config/
│       ├── core/
│       ├── stages/
│       └── utils/
├── tests/
│   ├── edge_cases/
│   ├── test_load.py
│   ├── test_pipeline_smoke.py
│   └── ...
├── LICENSE
├── README.md
└── pyproject.toml

---

🛠 Tech Stack

• Python 3.12+
• Scikit-learn
• Pandas & NumPy
• Optuna
• SHAP
• Typer + Rich
• Joblib
• Plotly

---

📄 License

This project is licensed under the MIT License.

---