Performance Tips¶

Optimize training speed and resource usage.

Speed Optimization¶

1. Reduce Population Size¶

# Faster (30-50 individuals)
ga:
  population_size: 40
  n_generations: 30

# Slower but better results (80-120 individuals)
ga:
  population_size: 100
  n_generations: 50

2. Reduce Generations¶

Monitor convergence and stop early if fitness plateaus.

3. Use Smaller Trees¶

tree:
  max_depth: 4        # Instead of 6
  min_samples_split: 15  # Instead of 8

4. Fewer CV Folds¶

experiment:
  cv_folds: 5  # Instead of 20 for quick experiments

Memory Optimization¶

Large Datasets¶

# Sample data for faster iteration
from sklearn.model_selection import train_test_split

# Use 20% of data for development
X_sample, _, y_sample, _ = train_test_split(X, y, train_size=0.2, stratify=y)

Large Populations¶

Monitor memory usage and reduce if needed:

# Monitor during training
watch -n 1 free -h

Parallel Fitness Evaluation¶

Future feature - currently sequential.