A bug? Strange behaviour with shape

[iago-pssjd]

Hi!

1. When summarising a multinomial model and placing model in the LHS of shape, output adds seemingly extra rows related to an inexistent group model.

library(nnet)

dat_multinom <- mtcars
dat_multinom$cyl <- sprintf("Cyl: %s", dat_multinom$cyl)

mod <- list(
    nnet::multinom(cyl ~ mpg, data = dat_multinom, trace = FALSE),
    nnet::multinom(cyl ~ mpg + drat, data = dat_multinom, trace = FALSE))

modelsummary(mod, shape = model + term + response ~ statistic, output = "markdown")

		response	Est.	S.E.
group	(Intercept)	Cyl: 6
		Cyl: 8
	mpg	Cyl: 6
		Cyl: 8
	drat	Cyl: 6
		Cyl: 8
(1)	(Intercept)	Cyl: 6	47.252	34.975
		Cyl: 8	72.440	37.175
	mpg	Cyl: 6	-2.205	1.638
		Cyl: 8	-3.580	1.775
	drat	Cyl: 6
		Cyl: 8
(2)	(Intercept)	Cyl: 6	89.573	86.884
		Cyl: 8	117.971	87.998
	mpg	Cyl: 6	-3.627	3.869
		Cyl: 8	-4.838	3.915
	drat	Cyl: 6	-3.210	3.810
		Cyl: 8	-5.028	4.199

The 6 rows relative to the group model (2nd to 7th) should not appear.

Thank you!

[iago-pssjd]

A second issue: for multinomial models coef_rename = TRUE only works for first response level.

library(nnet)
dat_multinom <- mtcars
dat_multinom <- transform(mtcars, vs = as.factor(vs), carb = as.factor(carb))
mod <- list(
    nnet::multinom(cyl ~ vs, data = dat_multinom, trace = FALSE),
    nnet::multinom(cyl ~ vs + carb, data = dat_multinom, trace = FALSE))
modelsummary(mod, shape = model + term + response ~ statistic, output = "markdown", coef_rename = TRUE)

		response	Est.	S.E.
group	(Intercept)	6
		8
	vs [1]	6
	vs1	8
	carb [2]	6
	carb [3]	6
	carb [4]	6
	carb [6]	6
	carb [8]	6
	carb2	8
	carb3	8
	carb4	8
	carb6	8
	carb8	8
(1)	(Intercept)	6	1.098	1.155
		8	2.639	1.035
	vs [1]	6	-2.015	1.297
	vs1	8	-13.237	63.322
	carb [2]	6
	carb [3]	6
	carb [4]	6
	carb [6]	6
	carb [8]	6
	carb2	8
	carb3	8
	carb4	8
	carb6	8
	carb8	8
(2)	(Intercept)	6	82.903	94.629
		8	82.866	18.933
	vs [1]	6	-83.819	94.630
	vs1	8	-184.305	0.000
	carb [2]	6	-94.450	227.108
	carb [3]	6	-16.791	0.000
	carb [4]	6	144.977	37.852
	carb [6]	6	258.280	0.000
	carb [8]	6	-27.108	0.000
	carb2	8	-81.480	18.936
	carb3	8	83.106	0.000
	carb4	8	146.112	37.852
	carb6	8	-106.938	0.000
	carb8	8	125.316	0.000

[vincentarelbundock]

Thanks a lot for the report. Both issues should be fixed on Github now.