tag:blogger.com,1999:blog-5652924.post7278466785776431374..comments2024-03-02T01:02:21.655-08:00Comments on Applied Data Science and <br>Machine Learning: Poll: Model SelectionDean Abbotthttp://www.blogger.com/profile/16818000233889520746noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-5652924.post-8850023058984608402007-03-03T20:37:00.000-08:002007-03-03T20:37:00.000-08:00For risk scoring it's nearly always KS (Kolmogorov...For risk scoring it's nearly always KS (Kolmogorov-Smirnov). For forecasting, MAPE.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-5652924.post-73366567134647753182007-02-06T06:34:00.000-08:002007-02-06T06:34:00.000-08:00For classification problems, my usual practice is ...For classification problems, my usual practice is to assess class separation via AUC (area under the ROC curve) and probability accuracy via informational loss or MSE.<br /><br />For numeric problems, I lean toward mean absolute error, or, when relative error is more important, mean absolute percent error.Will Dwinnellhttps://www.blogger.com/profile/03379859054257561952noreply@blogger.com