tag:blogger.com,1999:blog-5652924.post7620866915265496970..comments2024-03-02T01:02:21.655-08:00Comments on Applied Data Science and <br>Machine Learning: What To Take Home from Your Next Predictive Analytics ConferenceDean Abbotthttp://www.blogger.com/profile/16818000233889520746noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-5652924.post-50282628630209648612013-02-16T07:43:37.004-08:002013-02-16T07:43:37.004-08:00I have found it much easier to report failures to ...I have found it much easier to report failures to clients than at a conference. One problem is that there are so many failures, we could easily fill our time slots with them! <br />Dean Abbotthttps://www.blogger.com/profile/16818000233889520746noreply@blogger.comtag:blogger.com,1999:blog-5652924.post-44224216585380422782013-02-16T07:32:05.997-08:002013-02-16T07:32:05.997-08:00Good points, Dean. Statistical significance is a ...Good points, Dean. Statistical significance is a vary subtle issue, and we see yet again that the effect of multiple testing (especially when "multiple" implies a very large number of analyses) is potentially quite dangerous.<br /><br />Partly this happens because, as you say, there is an aversion to reporting "failure". During a recent conversation with a client, I Will Dwinnellhttps://www.blogger.com/profile/03379859054257561952noreply@blogger.comtag:blogger.com,1999:blog-5652924.post-11582845872872753472013-02-14T11:41:21.647-08:002013-02-14T11:41:21.647-08:00Very clear thinking. There is tremendous pressure ...Very clear thinking. There is tremendous pressure to be "successful" defined as accepting the alternative hypothesis, such that no, no change, and no difference are simply not perceived as valuable, especially if the customer discounts the knowledge gained from the process itself. Thank you.Anonymoushttps://www.blogger.com/profile/15943490251021380830noreply@blogger.com