This lecture emphasizes the domain of problems suited to data mining and the key problem of over-fitting.
Stepwise regression is a powerful data mining tool, if used carefully. This lecture also covers the 5 C's of data mining. The application for this class is diagnosing osteoporosis. Also, this JMP script animates the role of the thresholds that determine the costs of a classification rule.
This lecture might be skipped, depending on how things go in terms of what is covered in the first two lectures and the interests of the class. It covers logistic regression and related issues of calibration.
This lecture gets into a bit of logistic regression, stressing the important role of calibration. The lecture also introduces the tree-based models (CART). The application used in this lecture considers how to find the best candidates for a special training program.
As a technical supplement for those who are inclined, this handout describes likelihoods and how they show up in statistical modeling.
In addition to showing how trees work, this lecture shows how they can be used with regression to make a powerful combination. The lecture also touches on neural networks as yet another methodology.
My paper "An Introduction to Bootstrap Methods" (which appeared in Sociological Methods & Research back in 1989) introduces you to the ideas of bootstrap resampling through a variety of examples. The paper includes examples in regression and illustrates situations in which the bootstrap does not give the answer you'd like.