Energy Informatics

Table 1 Hyper-parameters possible and optimal values

From: Climatization and luminosity optimization of buildings using genetic algorithm, random forest, and regression models

Hyper-parameter	Possible Values	Optimal Value
Number of Trees	200 to 2000	400
Max Features	Auto, Sqrt or Log2	Sqrt
Criterion	Gini or Entropy	Gini
Max Depth	10 to 32	26
Min Samples to Split	2, 5 or 10	10
Min Samples in a Leaf	1,2 or 4	1
Bootstrap	True or False	False

Back to article page