Is the permuation-resampling method too conservative?

Is the permuation-resampling method too conservative?

However, often test resutls are strongly correlated (e.g. genes in the same pathway behave similalry) and the Bonferroni will be too conservative. However the permuation-resampling method still works in the presence of correaltions. Bonferroni tells us none of the adjusted p-values are significant, which we know is the wrong answer.

Which is an example of a resampling method?

We will make up some data – a tpical example is trying to identify genes that are differentially expressed in two groups of people, pehraps those who are helathy and those who are sick. For each gene, we can perform a t-test to see if the gene is differnetially expressed across the two groups at some nominal significanc level, typically 0.05.

How to test an appropirate statistic in Python?

An appropirate test statistic is the difference between the 7th percentile, and if we knew the null distribution of this statisic, we could test for the null hypothesis that the statistic = 0. Permuting the labels of the 2 data sets allows us to create the empirical null distribution.

How is permuatation resampling used in computational statistics?

Permuatation resampling is used ot generate the null distribtuion of labeled data by switching lebals. Because the number of permuations grows so fast, it is typically only feasible to use a Monte Carlo sample of the possible set of permuations in computation.

Which is the most efficient method for resampling?

This can be done using Boolean indexing as shown in the examples below, and is efficient since the operations are on views of the origianl array rather thna copies. LOOCV also uses the same idiom, and a simple example of LOOCV for model selection is illustrated.

How is permuatation resampling used to calculate Cook’s distance?

The calculation of Cook’s distance involves the fitting of n regression models, so we want to do this as efficiently as possible. Permuatation resampling is used ot generate the null distribtuion of labeled data by switching lebals.