tag:blogger.com,1999:blog-3366935554564939610.post7609930347080658499..comments2020-02-20T03:05:59.043-05:00Comments on Data Miners Blog: Three QuestionsMichael J. A. Berryhttp://www.blogger.com/profile/06077102677195066016noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-3366935554564939610.post-17621165122271943822008-09-30T01:11:00.000-04:002008-09-30T01:11:00.000-04:001. perfectly clear. thanks!2. my 2nd question has ...1. perfectly clear. thanks!<BR/><BR/>2. my 2nd question has more to do with impossible values. the example i gave was that of MPG. my research told me that an MPG above 60 for a fleet vehicle is very very unlikely. comparing this finding/assumption with the distribution (percentiles) of MPG in the dataset led to the conclusion that those values outside the range can only be data entry errors.<BR/>i agree that discarding values in the tails of the distribution can't be applied all the time.<BR/><BR/>3. thanks again for the explanation!!!<BR/><BR/>I just bought your book - "Data Mining Techniques, 2nd edition" last month, and I'm finding it very very good!!!Datalligencehttps://www.blogger.com/profile/16461960582799657275noreply@blogger.com