tag:blogger.com,1999:blog-590043897961646114.post1677344264188629686..comments2024-02-06T02:06:06.364-08:00Comments on Engaging Market Research: Regression with Multicollinearity Yields Multiple Sets of Equally Good CoefficientsUnknownnoreply@blogger.comBlogger2125tag:blogger.com,1999:blog-590043897961646114.post-3693802226775688052015-07-09T12:49:44.211-07:002015-07-09T12:49:44.211-07:00Any reader interested in the distinction between p...Any reader interested in the distinction between prediction and explanation should read the paper by Galit Shmueli. The article “To Explain or To Predict?” and an accompanying video can be found at her website, http://www.galitshmueli.com/.Joelhttps://www.blogger.com/profile/13462586727133296404noreply@blogger.comtag:blogger.com,1999:blog-590043897961646114.post-30909617445018884292015-07-09T11:26:55.673-07:002015-07-09T11:26:55.673-07:00Which "none" are you speaking of when yo...Which "none" are you speaking of when you say 'none of this is an issue for prediction'? I would argue multi-colinearity remains an issue in prediction because it's presence forces the new data to have more of the same structure than if you drop the collinear variables. The risks in deploying a model with multi-colinearity are higher, so its better to remove the variables with low information gain.Chrisnoreply@blogger.com