In the last week or so lm() in R-devel (and Rpatched, see
https://cran.rstudio.com/web/checks/check_results_pcts.html) started to produce
different names for some interactions, compared to R-3.6.x. A full example is
at the end of this email but the gist is that a term
"TimeIndex:Season", where "Season" is a factor produces names like
"TimeIndex:Season1", while R-devel changes the order of the names to
"Season1:TimeIndex", etc.
Is this change on purpose?
Georgi Boshnakov
---
ex_fac <- as.factor(rep(1:4, 10))
ex_fo <- as.formula("x ~ -1 - Season + TimeIndex:Season")
ex_data <- data.frame(x = rnorm(40), Season = ex_fac, TimeIndex = 1:40)
lm(ex_fo, data = ex_data, na.action = na.exclude)
## R version 3.6.1 (2019-07-05) -- "Action of the Toes"
##
## Call:
## lm(formula = ex_fo, data = ex_data, na.action = na.exclude)
##
## Coefficients:
## TimeIndex:Season1 TimeIndex:Season2 TimeIndex:Season3 TimeIndex:Season4
## 0.0036876 -0.0033619 0.0211838 -0.0008042
## R Under development (unstable) (2020-02-12 r77798) -- "Unsuffered
Consequences"
##
## Call:
## lm(formula = ex_fo, data = ex_data, na.action = na.exclude)
##
## Coefficients:
## Season1:TimeIndex Season2:TimeIndex Season3:TimeIndex Season4:TimeIndex
## -0.01033 0.02900 0.01791 0.01596
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
