So basically I'm using klaR to perform a greedy.wilks on a data frame that I obtained through the HCPC function from factomineR.
Here is my starting data frame (I'm only interested in 1 to 15 variables here) :
> alim[,1:15]
# A tibble: 101 x 15
poulet boeuf porc poisson `fruits de mer` `légume vert` `légume racine` `autre légume` féculent crudités pâtisseries `encas sucré` `encas salé` soda alcool
<dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 3.89 11.8 13.3 2.07 1.76 0.290 14.3 1.94 19.4 7.14 2.6 8.77 5.89 15.1 10.6
2 6.33 4.75 2.94 11.7 5.03 13.4 11.4 3.1 14.6 16.8 3.02 2.43 0.76 7.86 15.7
3 8.97 2.79 2.56 1.15 4.31 19.1 16.9 1.89 22.6 2.8 8.21 0.81 10.7 9.53 6.64
4 4.19 6.28 7.08 2.98 13.4 12.3 21.2 1.76 10.8 8.51 3.83 7.3 0.63 10.5 1.91
5 8.33 3.83 6.61 0.290 2.33 4.73 8.48 2.26 29.9 4.12 7.7 1.81 2.09 4.87 13.2
6 4.62 4.48 7.86 11.8 8.13 8.31 21.0 0.81 9.35 7.34 0.46 2.44 5.69 0.7 3.77
7 0.17 0.79 3.52 5.37 7.04 10.5 29.9 10.9 9.36 17.2 1.28 4.73 2.64 10.6 1.11
8 8.84 6.77 4.95 0.65 3.22 19.7 17.5 19.8 16.2 4.65 4.93 3.76 5.36 3.82 6.77
9 1.3 6.8 8.08 1.81 3.3 3.15 24.6 1.5 17.9 0.83 0.44 6.67 3.34 4.55 0.36
10 10.3 4.79 6.51 4.1 6.16 12.2 21.6 3.9 8.63 16.8 12.8 7.01 1.62 0.51 10.1
# … with 91 more rows
I'm now using PCA and HCPC from factomineR :
> alim_pca <- PCA(alim[,1:15], scale.unit = TRUE,graph = FALSE)
> alim_hcpc <- HCPC(alim_pca, nb.clust = 3, graph=FALSE, nb.par = Inf)
And I get a new data frame with a cluster column calculated from HCPC :
> alim_types <- alim_hcpc$data.clust
> alim_types
poulet boeuf porc poisson fruits de mer légume vert légume racine autre légume féculent crudités pâtisseries encas sucré encas salé soda alcool clust
1 3.89 11.85 13.32 2.07 1.76 0.29 14.31 1.94 19.40 7.14 2.60 8.77 5.89 15.06 10.56 1
2 6.33 4.75 2.94 11.70 5.03 13.41 11.43 3.10 14.63 16.75 3.02 2.43 0.76 7.86 15.66 2
3 8.97 2.79 2.56 1.15 4.31 19.07 16.92 1.89 22.60 2.80 8.21 0.81 10.70 9.53 6.64 1
4 4.19 6.28 7.08 2.98 13.37 12.28 21.21 1.76 10.84 8.51 3.83 7.30 0.63 10.46 1.91 2
5 8.33 3.83 6.61 0.29 2.33 4.73 8.48 2.26 29.86 4.12 7.70 1.81 2.09 4.87 13.22 1
6 4.62 4.48 7.86 11.77 8.13 8.31 20.98 0.81 9.35 7.34 0.46 2.44 5.69 0.70 3.77 2
7 0.17 0.79 3.52 5.37 7.04 10.49 29.94 10.88 9.36 17.21 1.28 4.73 2.64 10.55 1.11 2
8 8.84 6.77 4.95 0.65 3.22 19.73 17.53 19.82 16.23 4.65 4.93 3.76 5.36 3.82 6.77 2
9 1.30 6.80 8.08 1.81 3.30 3.15 24.58 1.50 17.94 0.83 0.44 6.67 3.34 4.55 0.36 1
10 10.34 4.79 6.51 4.10 6.16 12.22 21.57 3.90 8.63 16.85 12.77 7.01 1.62 0.51 10.12 2
Now I want to use greedy.wilks, I tried both "default" and "formula" usage but they both return errors :
> alim_greed <- greedy.wilks(alim_types[,1:15], alim_types$clust)
Error in parse(text = x, keep.source = FALSE) :
<text>:1:36: unexpected symbol
1: alim_types$clust ~ féculent+légume racine
^
> alim_greed <- greedy.wilks(clust ~ ., data = alim_types)
Error in `[.data.frame`(m, xvars) : undefined columns selected
I might be wrong on something but I can't find where.