ML Intro

Python Machine Learning Intro: Workflow & Ecosystem

Machine Learning (ML) in Python utilizes a rich scientific software stack (NumPy, Pandas, SciPy, Scikit-Learn, Matplotlib) to build predictive models from observational data.

Machine Learning Pipeline

flowchart LR
    A["Raw Data Collection"] --> B["Data Cleaning & Preprocessing"]
    B --> C["Feature Scaling & Train/Test Split"]
    C --> D["Model Selection & Training"]
    D --> E["Evaluation & Metrics"]

Learning Paradigm Categories

1. Supervised Learning: Training data includes input features ($X$) and target labels ($y$) (e.g. Classification, Linear Regression).
2. Unsupervised Learning: Discovering hidden patterns or clusters in unlabeled data (e.g. K-Means Clustering, PCA).
3. Reinforcement Learning: Agent learning actions via reward and penalty signals.

Practical Code Example

import numpy as np
from typing import Dict, Any

def demonstrate_ml_data_structure() -> Dict[str, Any]:
    """Construct a standard feature matrix (X) and target vector (y)."""
    # Feature Matrix X: 4 samples, 2 features (Age, Income)
    X = np.array([
        [25, 50000],
        [30, 60000],
        [35, 75000],
        [40, 90000]
    ], dtype=np.float64)

    # Target Vector y: Binary classification (0: No Purchase, 1: Purchase)
    y = np.array([0, 0, 1, 1], dtype=np.int32)

    return {
        "num_samples": X.shape[0],
        "num_features": X.shape[1],
        "X_mean": np.mean(X, axis=0),
        "y_labels": y
    }

if __name__ == "__main__":
    info = demonstrate_ml_data_structure()
    print(f"Feature Matrix Samples: {info['num_samples']}, Features: {info['num_features']}")
    print(f"Feature Means (Age, Income): {info['X_mean']}")

Best Practices & Gotchas

• Keep Data Matrices 2D: Feature matrices ($X$) must always be 2D arrays of shape (n_samples, n_features). Target vectors ($y$) are typically 1D of shape (n_samples,).
• Prevent Data Leakage: Fit transformers and scalers ONLY on training data, never on the combined dataset prior to splitting.
• Use NumPy Vectorization: Avoid looping over arrays in Python; use NumPy vectorized matrix math for speed.

Self-Check Challenge

Create a 2D NumPy feature array $X$ of shape (5, 3) with random float values and print its shape and mean.