False Discovery Rate

First published: 2005

Brief summary

Teaching PDF on multiple testing and false discovery rate.

Article

False Discovery Rate is covered by the linked source. Teaching PDF on multiple testing and false discovery rate.

The material from UC Berkeley describes a method for detecting, decomposing, testing or reproducing structure in time-series data.

Reliable use requires attention to sampling, trend, missing values, noise, boundaries, parameter choices and multiple testing. Sensitivity checks help assess stability.

No method can confirm a genuine cycle from an attractive graph or isolated peak alone. Significance requires an appropriate null model and documented preprocessing.

Source details and credits

• Source / publisher: UC Berkeley
• Source type: Teaching PDF
• URL type: PDF
• Credits: UC Berkeley
• URL:

  STAT.BERKELEY.EDU "https://www.stat.berkeley.edu/~mgoldman/Section0402.pdf"