In krisrs1128/mvarVis: Visualize multivariate analysis results.

This is an example from MFA() in the FactoMineR package, but done using our wrapper. The first two columns are supplementary variables.

library("mvarVis")
library("FactoMineR")
data(wine)
wine_mfa <- ordi(wine[, -c(1:2)], "MFA", wine[, 1:2], group = c(5,3,10,9,2),
                 type = rep("s",5), ncp = 3, graph = F)

We can plot the results with d3. Hover over points to see their values in the table that appears below the plot. Change the colors using the selection box at the top left.

plot_mvar_d3(wine_mfa)

If we want, we can have as many panels as we want. This splits the MFA variables by group. The splitting code isn't pretty, but it shouldnt' be too hard to write some helpers to streamline things.

split_layer <- function(x, start, finish) {
  new("mvarLayer", coord = x@coord[start:finish, ],
      annotation = x@annotation[start:finish, ])
}

wine_quanti <- wine_mfa@table[["quanti.var"]]
wine_mfa_split <- new(
  "mvarTable",
  table = list(
    new("mvarLayer", coord = wine_mfa@table[["ind"]]@coord,
        annotation = wine_mfa@table[["ind"]]@annotation),
    split_layer(wine_quanti, 1, 5),
    split_layer(wine_quanti, 6, 8),
    split_layer(wine_quanti, 9, 18),
    split_layer(wine_quanti, 19, 27),
    split_layer(wine_quanti, 28, 29)
  ),
  eig = wine_mfa@eig
)

plot_mvar_d3(wine_mfa_split, c(list(list("point")), replicate(5, list("text", "arrow"), simplify = F)))

We can also have just one panel.

wine_pca <- ordi(wine[, -c(1:2)], "factominer_pca",
                 table_names = "ind",
                 ncp = 3, graph = F)
plot_mvar_d3(wine_pca)

Here's an MCA example. Notice we can customize width / height.

data(hobbies)
hobbies_mca <- ordi(hobbies[, 1:18], method = "MCA",
                    rows_annot = hobbies[, 19:23], graph = F, ncp = 3)

plot_mvar_d3(hobbies_mca, height = 400, rMax = 8)

krisrs1128/mvarVis documentation built on Oct. 13, 2019, 11:14 p.m.