ml-model-training

name	ML Model Training
description	Build and train machine learning models using scikit-learn, PyTorch, and TensorFlow for classification, regression, and clustering tasks

ML Model Training

Training machine learning models involves selecting appropriate algorithms, preparing data, and optimizing model parameters to achieve strong predictive performance.

Training Phases

• Data Preparation: Cleaning, encoding, normalization
• Feature Engineering: Creating meaningful features
• Model Selection: Choosing appropriate algorithms
• Hyperparameter Tuning: Optimizing model settings
• Validation: Cross-validation and evaluation metrics
• Deployment: Preparing models for production

Common Algorithms

• Regression: Linear, Ridge, Lasso, Random Forest
• Classification: Logistic, SVM, Random Forest, Gradient Boosting
• Clustering: K-Means, DBSCAN, Hierarchical
• Neural Networks: MLPs, CNNs, RNNs, Transformers

Python Implementation

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split, cross_val_score
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import (accuracy_score, precision_score, recall_score,
                            f1_score, confusion_matrix, roc_auc_score)
import torch
import torch.nn as nn
from torch.utils.data import DataLoader, TensorDataset
import tensorflow as tf
from tensorflow import keras

# 1. Generate synthetic dataset
np.random.seed(42)
n_samples = 1000
n_features = 20

X = np.random.randn(n_samples, n_features)
y = (X[:, 0] + X[:, 1] - X[:, 2] + np.random.randn(n_samples) * 0.5 > 0).astype(int)

# Split data
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)

# Normalize features
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

print("Dataset shapes:")
print(f"Training: {X_train_scaled.shape}, Testing: {X_test_scaled.shape}")
print(f"Class distribution: {np.bincount(y_train)}")

# 2. Scikit-learn models
print("\n=== Scikit-learn Models ===")

models = {
    'Logistic Regression': LogisticRegression(max_iter=1000),
    'Random Forest': RandomForestClassifier(n_estimators=100, random_state=42),
    'Gradient Boosting': GradientBoostingClassifier(n_estimators=100, random_state=42),
}

sklearn_results = {}
for name, model in models.items():
    model.fit(X_train_scaled, y_train)
    y_pred = model.predict(X_test_scaled)
    y_pred_proba = model.predict_proba(X_test_scaled)[:, 1]

    sklearn_results[name] = {
        'accuracy': accuracy_score(y_test, y_pred),
        'precision': precision_score(y_test, y_pred),
        'recall': recall_score(y_test, y_pred),
        'f1': f1_score(y_test, y_pred),
        'roc_auc': roc_auc_score(y_test, y_pred_proba)
    }

    print(f"\n{name}:")
    for metric, value in sklearn_results[name].items():
        print(f"  {metric}: {value:.4f}")

# 3. PyTorch neural network
print("\n=== PyTorch Model ===")

class NeuralNetPyTorch(nn.Module):
    def __init__(self, input_size):
        super().__init__()
        self.fc1 = nn.Linear(input_size, 64)
        self.fc2 = nn.Linear(64, 32)
        self.fc3 = nn.Linear(32, 1)
        self.relu = nn.ReLU()
        self.dropout = nn.Dropout(0.3)

    def forward(self, x):
        x = self.relu(self.fc1(x))
        x = self.dropout(x)
        x = self.relu(self.fc2(x))
        x = self.dropout(x)
        x = torch.sigmoid(self.fc3(x))
        return x

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
pytorch_model = NeuralNetPyTorch(n_features).to(device)
criterion = nn.BCELoss()
optimizer = torch.optim.Adam(pytorch_model.parameters(), lr=0.001)

# Create data loaders
train_dataset = TensorDataset(torch.FloatTensor(X_train_scaled),
                             torch.FloatTensor(y_train).unsqueeze(1))
train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)

# Train PyTorch model
epochs = 50
pytorch_losses = []
for epoch in range(epochs):
    total_loss = 0
    for batch_X, batch_y in train_loader:
        batch_X, batch_y = batch_X.to(device), batch_y.to(device)

        optimizer.zero_grad()
        outputs = pytorch_model(batch_X)
        loss = criterion(outputs, batch_y)
        loss.backward()
        optimizer.step()
        total_loss += loss.item()

    pytorch_losses.append(total_loss / len(train_loader))
    if (epoch + 1) % 10 == 0:
        print(f"Epoch {epoch + 1}/{epochs}, Loss: {pytorch_losses[-1]:.4f}")

# Evaluate PyTorch
pytorch_model.eval()
with torch.no_grad():
    y_pred_pytorch = pytorch_model(torch.FloatTensor(X_test_scaled).to(device))
    y_pred_pytorch = (y_pred_pytorch.cpu().numpy() > 0.5).astype(int).flatten()
    print(f"\nPyTorch Accuracy: {accuracy_score(y_test, y_pred_pytorch):.4f}")

# 4. TensorFlow/Keras model
print("\n=== TensorFlow/Keras Model ===")

tf_model = keras.Sequential([
    keras.layers.Dense(64, activation='relu', input_shape=(n_features,)),
    keras.layers.Dropout(0.3),
    keras.layers.Dense(32, activation='relu'),
    keras.layers.Dropout(0.3),
    keras.layers.Dense(1, activation='sigmoid')
])

tf_model.compile(
    optimizer='adam',
    loss='binary_crossentropy',
    metrics=['accuracy']
)

history = tf_model.fit(
    X_train_scaled, y_train,
    batch_size=32,
    epochs=50,
    validation_split=0.2,
    verbose=0
)

y_pred_tf = (tf_model.predict(X_test_scaled) > 0.5).astype(int).flatten()
print(f"TensorFlow Accuracy: {accuracy_score(y_test, y_pred_tf):.4f}")

# 5. Visualization
fig, axes = plt.subplots(2, 2, figsize=(12, 10))

# Model comparison
models_names = list(sklearn_results.keys()) + ['PyTorch', 'TensorFlow']
accuracies = [sklearn_results[m]['accuracy'] for m in sklearn_results.keys()] + \
             [accuracy_score(y_test, y_pred_pytorch),
              accuracy_score(y_test, y_pred_tf)]

axes[0, 0].bar(range(len(models_names)), accuracies, color='steelblue')
axes[0, 0].set_xticks(range(len(models_names)))
axes[0, 0].set_xticklabels(models_names, rotation=45)
axes[0, 0].set_ylabel('Accuracy')
axes[0, 0].set_title('Model Comparison')
axes[0, 0].set_ylim([0, 1])

# Training loss curves
axes[0, 1].plot(pytorch_losses, label='PyTorch', linewidth=2)
axes[0, 1].plot(history.history['loss'], label='TensorFlow', linewidth=2)
axes[0, 1].set_xlabel('Epoch')
axes[0, 1].set_ylabel('Loss')
axes[0, 1].set_title('Training Loss Comparison')
axes[0, 1].legend()
axes[0, 1].grid(True, alpha=0.3)

# Scikit-learn metrics
metrics = ['accuracy', 'precision', 'recall', 'f1']
rf_metrics = [sklearn_results['Random Forest'][m] for m in metrics]
axes[1, 0].bar(metrics, rf_metrics, color='coral')
axes[1, 0].set_ylabel('Score')
axes[1, 0].set_title('Random Forest Metrics')
axes[1, 0].set_ylim([0, 1])

# Validation accuracy over epochs
axes[1, 1].plot(history.history['accuracy'], label='Training', linewidth=2)
axes[1, 1].plot(history.history['val_accuracy'], label='Validation', linewidth=2)
axes[1, 1].set_xlabel('Epoch')
axes[1, 1].set_ylabel('Accuracy')
axes[1, 1].set_title('TensorFlow Training History')
axes[1, 1].legend()
axes[1, 1].grid(True, alpha=0.3)

plt.tight_layout()
plt.savefig('model_training_comparison.png', dpi=100, bbox_inches='tight')
print("\nVisualization saved as 'model_training_comparison.png'")

print("\nModel training completed!")

Training Best Practices

• Data Split: 70/15/15 for train/validation/test
• Scaling: Normalize features before training
• Cross-validation: Use K-fold for robust evaluation
• Early Stopping: Prevent overfitting
• Class Balancing: Handle imbalanced datasets

Key Metrics

• Accuracy: Overall correctness
• Precision: Positive prediction accuracy
• Recall: True positive detection rate
• F1 Score: Harmonic mean of precision/recall
• ROC-AUC: Threshold-independent metric

Deliverables

• Trained model checkpoint
• Performance metrics on test set
• Feature importance analysis
• Learning curves
• Hyperparameter configuration
• Model evaluation report