Random Search Method

What is Random Search?

Random Search samples hyperparameter combinations randomly from specified distributions or ranges.

How It Works

  1. Define parameter ranges or distributions
  2. Randomly sample N combinations
  3. Evaluate with cross-validation
  4. Return best combination found

Advantages

  • Faster: Fixed number of iterations
  • Better Coverage: Can explore more of the space
  • Good Enough: Often finds near-optimal solutions

Disadvantages

  • Not Guaranteed: May miss the absolute best
  • Randomness: Results vary between runs

When to Use

  • Large parameter spaces
  • Limited computational budget
  • When good (not perfect) is acceptable

Related Terms

  • Grid Search: Exhaustive alternative
  • Hyperparameter Tuning: The broader process