- The unknown test set does not
classify similarly to the learning set,
indicating that the suggested partial classification
potential for category#2 data records
(fig.14A)
of the learning set represents a statistical aberration,
due to a slight difference of the category#1
and category#2 data records selected from the random data set.
The difference concerns 3 of the 134 data columns
(fig.15)
and is
statistically not significant.
- The test set patients #001,002,009,010,019,020,029,030,039,040
were labeled with: ?, and thus remained unknown to the learning process
as can be checked data record numbers < #26 in
fig.16
where these data records are missing.
- Displayed are the classification results for all test set
cat#1 and cat#2 data records.