Skip to main content

Table 2 Average performance (SCC) of weighted averaging models with different combinations of normalization functions and weighted methods

From: A hybrid framework for improving uncertainty quantification in deep learning-based QSAR regression modeling

Splitting strategies

Weighted methods

MinMax

Zscore

Rank

IVIT

Unweighted

0.306

0.315

0.314

Weighted

0.314

0.317

0.312

IVOT

Unweighted

0.230

0.236

0.240

Weighted

0.239

0.239

0.236

OVOT

Unweighted

0.200

0.207

0.211

Weighted

0.204

0.206

0.208

  1. Better results between the weighted and unweighted approaches for each combination of splitting strategy and normalization function are highlighted in boldface type
  2. A higher SCC indicates better performance