Classification rules
To discover classification rules for our dataset, starting from
angiogenesis-related SNPs and identified SNPs in ADME genes (sorafenib
dataset), we applied a classifier for the identification of rules which
could facilitate the stratification of responder/non-responder patients
in terms of sorafenib response. Classifiers belong to the non-parametric
supervised learning algorithm category, where the machines learn
patterns buried in the data using previously labeled training data which
is mandatory for supervised learning. Supervised learning aims to
produce a model that predicts class variable values by learning simple
patterns inferred from the data features. The RandomTree’s output is a
classification tree easily understandable even by non-domain experts
which can be quickly translated into classification rules in the ”IF
(cond1 & cond2 & … & condn) THEN class” format.