Table 5 Analysis of classification error of kNN model performance.

From: An expression signature at diagnosis to estimate prostate cancer patients’ overall survival

Model

Parameters [scale]

Training set (n=139) classification error (overall survival, years)

Test set (n=50) classification error (overall survival, years)

P-value (is random)

  

Average error

Standard deviation

Average error

Standard deviation

 

RND

Random numbers

  

3.85

3.18

 

1

Age, WHO, CS, log-PSA [1,9,3,9]

2.97

2.14

3.44

2.93

0.2300

2

IGFBP3, VGLL3, F3, min(3G), max(3G) [1,9,1,3,1]

2.82

2.36

3.14

2.71

0.0373

3

Age, log-PSA, IGFBP3, VGLL3, F3, min(3G), max(3G) [1,3,1,9,3,3,1]

2.85

2.05

2.69

2.34

0.0024

4

Age, WHO, CS, log-PSA, IGFBP3, VGLL3, F3, min(3G), max(3G) [1,9,9,3,1,9,1,3,1]

2.81

1.76

2.72

2.41

0.0038

  1. Abbreviations: CS, clinical stage; 3G, IGFBP3, VGLL3, F3; kNN, k-nearest neighbor; WHO, World Health Organization tumor grading.
Back to article page