qe functions do not return same structure of prediction objects

[nks1974]

When running a problem, in qeML v1.1, on many qe functions for comparison, the prediction objects do not have the same structure. This makes comparison more difficult.

Example with factor target and numeric x-vars:

• qeAdaBoost will return a prediction object with the structure containing predClasses and probs incl class.
• qeKNN will return predClasses and probs excl class.
• qeDT will return only predClasses
• qeKNNna will return only a numeric (probs but unnamed)

Would it be possible to align the output for easy comparison, eg to a named list like for qeAdaBoost? This would also make it easier to expand qeCompare.

[matloff]

You have a good point, and I'll see what I can do. Please note that full consistency is impossible, because some ML methods do not automatically compute the posterior class probabilities. SVM, for instance, does not do so, though some implementations add on "calibration"-computed probs.

[rdanroces]

Good day Mr. Matloff:

I recently purchased a pre-print version of your book, and I must say I am learning. a lot already. My problem is I haven't really been able to follow along the examples real time as I am having trouble installing the qeML package. Could you help?

These are the errors I'm getting:

Warning in install.packages :
package ‘qeML’ is not available for this version of R

A version of this package for your version of R might be available elsewhere,
see the ideas at
https://cran.r-project.org/doc/manuals/r-patched/R-admin.html#Installing-packages

Now I have tried via your github repository, regtools, and even remotes, to no avail. For reference I am using macbook air 2020 (last intel version and not M1 and M2), with R version 4.3.1, and RStudio version of Version 2023.06.0+421.

[matloff]

Thanks for using my book!

I had not realized that preprint versions were available. The qeML package is not yet on CRAN. Instead, use devtools (install first if you don't already have it): devtools::install_github('matloff/qeML')