Petrophysical Regression regarding Porosity, Permeability, and Water Saturation Driven by Logging-Based Ensemble and Transfer Learnings: A Case Study of Sandy-Mud Reservoirs

Table 4

Initial settings of light gradient boosting machine (LightGBM) predictor and optimizers, and variation ranges of hyper-parameters.

Initial setting of core predictor	^1; ^2; ^3; ^4
Initial setting of optimizer ^5	RS	^6; ^7
	PSO	^8
	Bayes	^9
Variation ranges of hyperparameters

¹CART = classification and regression tree; ²max bins employed by Histogram algorithm to split a leaf node; ³min leafs required at a leaf node or there will have a cut for this leaf node; ⁴min gain required to split a leaf node or the growth of this node will be ceased; ⁵RS = random search; PSO = particle swarm optimization; Bayes = Bayesian optimization; ⁶lower and upper limits set by a fifth and five times of target , respectively; ⁷lower and upper limits set by a twentieth and twenty times of log10 base of target , respectively; ⁸random values varying within [0,1]; ⁹GP = Gaussian process.