Towards Optimization of Boosting Models for Formation Lithology Identification

Table 3

Performance matrix of 5-fold cross validation over the AdaBoost classifier, Gradient Tree Boosting classifier, XGBoost classifier, and stacked classifier in the Daniudi gas field.

	Precision	Recall	f1 score
AdaBoost
C	0.996	0.935	0.964
CR	0.96	0.901	0.925
CS	0.629	0.585	0.603
FS	0.788	0.78	0.783
M	0.859	0.872	0.863
MS	0.779	0.835	0.805
PS	0.82	0.807	0.811
S	0.786	0.791	0.777
Avg	0.825	0.818	0.819
XGBoost
C	0.991	0.938	0.963
CR	0.941	0.957	0.947
CS	0.592	0.629	0.601
FS	0.758	0.782	0.767
M	0.855	0.862	0.857
MS	0.788	0.759	0.771
PS	0.793	0.796	0.791
S	0.762	0.74	0.74
Avg	0.81	0.801	0.802
Gradient Tree Boosting
C	0.969	0.961	0.964
CR	0.93	0.924	0.925
CS	0.614	0.605	0.602
FS	0.812	0.779	0.793
M	0.881	0.861	0.869
MS	0.781	0.815	0.793
PS	0.84	0.832	0.833
S	0.801	0.832	0.808
Avg	0.826	0.824	0.823
Stacking result
C	0.99	0.962	0.974
CR	0.94	0.949	0.943
CS	0.64	0.668	0.648
FS	0.819	0.79	0.802
M	0.888	0.874	0.879
MS	0.801	0.809	0.803
PS	0.839	0.82	0.826
S	0.798	0.858	0.819
Avg	0.841	0.833	0.832