Grid Search

noun

say it: “GRID SURCH” /ˈɡrɪd ˈsɝtʃ/

Finding good settings for a model by trying every combination from a list of values and keeping the one that scores best.

What is grid search?

Grid search tries every combination of the hyperparameter values you list, such as depths 2, 4 and 8 with leaf sizes 1, 5 and 10, scores each with cross-validation, and keeps the best. It's thorough but slow: each value you add multiplies the number of models trained.

In a sentence

  • “Adding a third setting made the grid search train 60 models instead of 12.”
  • “Grid search tried nine combinations of depth and leaf size.”

Level: Journeyman · More AI Basics terms at this level

Updated