Table 3 Mean classifier test and blind well results (using a 20-run average) for designed ensemble based on Mean. F and Mean. K (Percentage-wise).

From: An ensemble-based machine learning solution for imbalanced multiclass dataset during lithology log generation

Method

ensenble type

Adaptation

\({\mathrm{Mean}.\mathrm{ F}}_{\mathrm{t}}\)

\({\mathrm{Mean}.\mathrm{ F}}_{\mathrm{b}}\)

\({\mathrm{Rank}}_{\mathrm{b}}\)

\({\mathrm{Mean}.\mathrm{K}}_{\mathrm{t}}\)

\({\mathrm{Mean}.\mathrm{ K}}_{\mathrm{b}}\)

\({\mathrm{Rank}}_{\mathrm{b}}\)

ECOC

Enhanced weighted average ensemble of SVM and RF in soft voting mode

CSL

94.92

91.04

1

91.70

84.50

1

Enhanced weighted average ensemble of SVM and RF in hard voting mode

94.07

90.33

2

90.44

83.62

2

  1. The t-index signifies test grades, while the b-index denotes ratings from blind evaluations.
Back to article page