This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Statistics 512: Final Project Due December 14, 2009 PLEASE READ CAREFULLY Introduction The aim of many statistical techniques is to provide a method for making good predictions. For the purposes of this project, you are given six sets of data, each in two parts: a training set with predictors and response with which you are to build your model(s), and a test set with just predictor variables. The assignment is to make predictions using whatever techniques or models you think are appropriate. Your grade will be based in part on how close your predictions are to the (hidden) test data responses, which were manufactured using the same processes as those used to produce the training data. In addition to making good predictions, a good statistical technique will estimate how far off its predictions are expected to be. (We have often seen this value encapsulated in the residual variance ( σ 2 ) term.) The second part of this project involves coming up with a good estimate of the error you expect to make. In this case, you should attempt to predict the sum of squared residuals of your predictions for the test data. The underlying true models are all “near” the standard linear models we have discussed in class. It is certainly possible that some of the departures we have considered will be included. Don’t expect every deviation in every data set, and don’t expect that each data set should have exactly one of the deviations....
View Full Document
- Fall '08