Mastering Machine Learning Algorithms

Inductive learning

Contrary to transductive learning, inductive learning considers all the X samples and tries to determine a complete p(x|y) or a function y=f(x) that can map both labeled and unlabeled points to their corresponding labels. In general, this method is more complex and requires more computational time; therefore, according to Vapnik's principle, if not required or necessary, it's always better to pick the most pragmatic solution and, possibly, expand it if the problem requires further details.