Table 1 Average model quality characteristics obtained in 50 iterations of cross-validation trained and tested on initial data, on the data with randomly permuted sample names, and after substitution of response values with the response potential values in the testing sets.

From: The hallmarks of dietary intervention-resilient gut microbiome

Model

Caret R2a

Does the mean caret R2 increase while testing on response potential compared to the initial model?

Intervention

Initial model

Random model

Initial, tested on response potential

HFD

0.11 ± 0.07

0.01 ± 0.02

0.15 ± 0.08

yes

FDP

0.1 ± 0.09

0.02 ± 0.06

0.05 ± 0.06

no

F.POT

0.13 ± 0.15

0.04 ± 0.1

0.18 ± 0.16

yes

F.HIM

0.2 ± 0.16

0.00 ± 0.1

0.25 ± 0.18

yes

F.INU

0.45 ± 0.19

0.02 ± 0.06

0.35 ± 0.14

no

F.ACC

0.15 ± 0.19

0.04 ± 0.1

0.2 ± 0.17

yes

MAL

0.06 ± 0.16

0.00 ± 0.12

0.08 ± 0.17

yes

NM

0.09 ± 0.09

0.00 ± 0.04

0.17 ± 0.13

yes

  1. acalculated as it is proposed in the caret package [72]: the square of the Pearson correlation coefficient between the true values in the test set and the predicted values.
Back to article page