next up previous contents
Next: Some statistical techniques and Up: Basic Statistical references for Previous: Shrinking, regularization and Bayesian   Contents

Exploratory and confirmatory data analysis: the Bootstrap

Tuckey initially suggested the distinction between exploratory and confirmatory data analysis. The first consisting in ``finding patterns'' in data, the second one in attempting to validate them, making sure that the perceived association are ``real'' and not due to random chance. Because of the vast amount of search involved in the analysis of microarray data it is often quite difficult to use appropriate validation methods. The bootstrap, an ``assumption free'' technique can be particularly useful.

References



2001-09-19