Geofluids

Research Article

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

Table 5

Initial settings of competitive predictors and variation ranges of applied hyperparameters.
Core predictor 1Initial settingVariation ranges of hyperparameters
KNN
2
3
8
9
SVR4

5


RF6

7



1KNN = k-nearest neighbors; SVR = supper vector regression; FR = random forest; 2“1” means KD-tree, “2” means Ball-tree; 3“1” means Manhattan distance, “2” means Euclidean distance; 4an upper bound on the fraction of training errors and a lower bound of the fraction of support vectors; 5control the window length of each probability density distribution; 6CART = classification and regression tree; 7min samples required to split a leaf node or the growth of this node will be ceased; 8only 1 or 2 will be chosen during iteration; 9only 1 or 2 will be chosen during iteration; RBF = radial basis function, which is a non-hyperparameter.