LGBM-CBFS: A Heuristic Feature Sampling Method Based on Tree Ensembles

Algorithm 1

Contribution-based feature sampling algorithm.

	Algorithm SelectTopN
	Input: featImportances//the vector containing importance scores of valid features
	Input: validFeatIndices//the vector containing position indices of valid features
	Input: featCount//the number of valid features
(1)	//the result set
(2)	//sort features based on scores
(3)	Sort(validFeatIndices, featImportances)
(4)	validFeatIndices.size()feature_fraction,
(5)	Min(, featCount)
(6)	The first elements in
(7)
(8)
(9)
(10)	Return