The three methods of machine learning are: classification, clustering, and regression.

In out workshops at the Melbourne Data Science Week we spent most of our time on classification
First tip was straight out of the school of Aidan: visualise your model in the data space, rather than just visualising your data in the model space:
I understood part of the method of classification as learning that defining characteristics from a history (a sample data set with known categories) and create a map in the data space to differentiate between the categories.
more to add…