Using pdmclass

Classification is a statistical technique that uses measurements on a defined set of samples (a training set) to build a rule that can be used to infer the group membership of future samples. An example would be using gene expression data to classify cancer patients according to the expected response to a certain course of therapy. There are many ways to build classification rules, but the general idea is the same; find patterns in the training set that are unique to each sample type and use this information to determine the class of new samples. Microarrays hold great promise for building classifiers because of the amount of information that can be generated from each array. However, this is a two edged sword – much of the information cannot be distinguished from noise (low expressing genes), and having many more observations than samples can make the analysis computationally and statistically difficult. Therefore, it is usually desirable to pre-filter the genes down to a much smaller (100- 200) set of genes before building the classifier. This itself i