| NEWS | R Documentation |
Major user visible change: Tor Wentzel-Larsen gently pointed out that I misspelled Theil's name as "Thiel" more than 1/2 the time (but not always), including the actual name of the function. This has been repaired.
Update the documentation to acknowledge "total least squares", a completely separate, and larger, literature that parallels and extends Deming regression.
Add a further data check. One user used the deming 'cv' option along with data that was negative (which implies a negative variance), and the routine did not fail nicely.
Version 1.2 had left a browser() call behind, which was encountered when doing regression through the origin. Pointed out by Daniela Herold. Add a no intercept regression to the test suite.
Add importFrom directive for calls to functions from "stats", per 2015/7/1 change in CRAN checks
Fix bug for the case of no intercept and the presence of stdx and stdy: an "as.matrix" was needed for the lm.wfit call.
Change the bootstrap to use residuals orthagonal to the fitted line. This makes the functions rotationally invariant.
Add the id option to the deming function, to allow for group jackknife estimates of variance.