Hello collinearity aficionados, I'm doing some preliminary work on revamping gretl's BKW collinearity diagnostics, and would appreciate feedback from any interested parties.
In git and snapshots there's now an (undocumented) built-in function BKW() which returns a matrix and basically has the same signature as bkw() in Lee Adkins' function package of that name -- except that in BKW() the second (parameter-names) argument can be a string, an array of strings, or omitted. So far, so uninteresting. But I've also tried to build a guide to interpretation on top of that function -- based on Lee's paper on the subject at https://learneconometrics.com/pdf/Collin/collin_gretl.pdf Work in progress can be found in http://users.wfu.edu/cottrell/tmp/bkw_read.inp # "interpreter" http://users.wfu.edu/cottrell/tmp/bkw_enroll.inp # examples These require current git. Have I got this right? Could something like this be helpful? Allin
