r - Convert to tidy format and Plot four density plots in the same plot

Question

Suppose I have data where x, y, z are three different data generating methods, and the columns refer to beta_0, beta_1, beta_2, and beta_3. I want to convert this to tidy format and then plot the density of beta_0 in one plot (with 3 curves), ..., the density of beta_3 in one plot (with 3 curves), all four plots in one output plot (so in a 4x4 grid or 4x1 grid, whatever, possibly using facet_wrap?

My question is: what's the tidy format of this data, if it can be put in tidy format, and then the code to generate the four densities (with each plot having three density curves) using ggplot2?

Toy data (the "real" data has 200 observations for each method, not just 6):

x=matrix(c(4.790584, 2.654499, 1.3987716, 3.504921,
           4.126791, 2.280143, 1.1348528, 3.084783,
           4.555479, 2.226738, 1.7021107, 3.500337,
           4.954719, 2.592864, 1.2658565, 4.213174,
           3.608878, 2.059524, 0.5363011, 2.993644,
           4.399778, 2.001757, 1.0602126, 3.377298), nrow=6, ncol=4, byrow=TRUE)
y=matrix(c(6.850924, 3.694619, 2.549025, 6.045664,
7.740951, 4.504154, 2.513180, 6.255492,
5.026915, 2.540771, 1.726427, 4.743082,
4.835503, 2.842332, 1.666898, 4.308631,
8.186425, 5.331871, 2.919130, 6.724499,
4.641722, 2.646673, 1.486447, 3.931251), nrow=6, ncol=4, byrow=TRUE)
z=matrix(c(4.651634, 2.611540, 1.5819549, 4.107449,
3.716444, 2.097695, 1.0136387, 3.051602,
4.184544, 1.986585, 1.0731731, 3.442379,
6.309591, 2.720760, 2.1551536, 4.572733,
6.614165, 3.253548, 2.6207540, 5.157300,
4.408838, 1.920230, 0.9850924, 3.320052), nrow=6, ncol=4, byrow=TRUE)
tidy=cbind(x, y, z)

question from:https://stackoverflow.com/questions/65944385/convert-to-tidy-format-and-plot-four-density-plots-in-the-same-plot

Answer 1

One option is to use as.data.frame.table and map_dfr to convert the matrices into a long format "tidy" dataframe.

The first step is to give the matrices meaningful names. Several ways to do this, but one easy option is provideDimnames. Then feed the named matrices into map_dfr.

library(tidyverse)

# name the mats
mats_named <- list(x,y,z) %>%
  purrr::set_names(c('x','y','z')) %>%
  map(function(x) provideDimnames(x, base = list('row','beta'), sep= '_'))

#------
$x
          beta   beta_1    beta_2   beta_3
row   4.790584 2.654499 1.3987716 3.504921
row_1 4.126791 2.280143 1.1348528 3.084783
row_2 4.555479 2.226738 1.7021107 3.500337
row_3 4.954719 2.592864 1.2658565 4.213174
row_4 3.608878 2.059524 0.5363011 2.993644
row_5 4.399778 2.001757 1.0602126 3.377298

$y
          beta   beta_1   beta_2   beta_3
row   6.850924 3.694619 2.549025 6.045664
row_1 7.740951 4.504154 2.513180 6.255492
...

Convert the matrices into a long format df and make the figure.

mats_named %>%
  map_dfr(as.data.frame.table, .id = 'method') %>%
  ggplot() +
  geom_density(aes(Freq, fill = method), alpha = 0.5) +
  facet_wrap(~Var2)