Bucket evaluations

From Wikipedia, the free encyclopedia
Jump to: navigation, search

In statistics, bucket evaluations is a method for correlating vectors. This method is a non-parametric, unsupervised correlation method first published in 2012 by Shabtai et al.[1]

Bucket evaluations was initially constructed for genetic research, and was used for finding a new potential anti-cancer drug.[2]

Bucket evaluations is named after the technique used to compare vectors in a matrix. Values in the vector are compared in sections (buckets). The buckets are set in a descending order, where the smallest buckets hold the highest scores, and have the strongest effect on the final correlation score. The similarity between vectors is calculated by comparing the ranks of the scores in each bucket, which are summed up to a similarity score.


  1. ^ Shabtai, Daniel; Guri Giaever; Corey Nislow (25 September 2012). "An algorithm for chemical genomic profiling that minimizes batch effects: Bucket evaluations". BMC Bioinformatics. 13: 245. PMC 3780717Freely accessible. PMID 23009392. doi:10.1186/1471-2105-13-245. 
  2. ^ Cheung-Ong, Kahlin; Kyung Tae Song; Zhidong Ma; Daniel Shabtai; Anna Y. Lee; David Gallo; Lawrence E. Heisler; Grant W. Brown; Ulrich Bierbach; Guri Giaever; Corey Nislow (August 28, 2012). "Comparative Chemogenomics To Examine the Mechanism of Action of DNA-Targeted Platinum-Acridine Anticancer Agents". ACS Chem. Biol. 7 (11): 120905082953006. PMC 3500413Freely accessible. PMID 22928710. doi:10.1021/cb300320d.