Skip to main content
. 2005 Aug 26;6(9):R76. doi: 10.1186/gb-2005-6-9-r76

Table 3.

Errors of two-step machine learning model

Step one Error (absolute number)
All sarcoma (n = 96) 0.09 (9)

Step two

MSS and MPNST (n = 14) 0.14 (2)
NFB and SWN (n = 8) 0.125 (1)
ARMS and ERMS (n = 7) 0.14 (1)
LMA/WLS and MLS (n = 12) 0

Cross-validation errors for step one and two of our prediction model are shown. The algorithm focuses on these samples of each row in isolation. Note that all the samples wrongly classified in step two were correctly classified in step one.