tdunning commented on PR #18103: URL: https://github.com/apache/pinot/pull/18103#issuecomment-4195203270
The algorithmic changes between 3.2 and 3.3 which were intended to provide a firm bound on the number of centroids had the side effect of changing the relationship between the number of centroids and the compression factor by about a factor of 2:1 (old:new). The accuracy is primarily driven by the number of retained centroids. This means that you have to have a higher compression parameter to keep the same number of centroids to keep the same level of accuracy. I am surprised, however, by the assertion that a 5:1 increase in compression parameter is required and would want to understand more about the test that supports that assertion. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
