Group-wise color in scatterplot

[araikes]

Hello,
First, thank you for your awesome product. I really appreciate the data exploration tool you've put together.

I'm trying to figure out if certain functionality exists. I'm running DataExplorer v 0.8.0.

I have the following dataframe:

Classes ‘tbl_df’, ‘tbl’ and 'data.frame':	32 obs. of  7 variables:
 $ group          : Factor w/ 2 levels "amber","blue": 1 2 2 1 2 1 1 2 2 2 ...
 $ ess_score      : num  13 14 9 12 1 3 11 8 5 8 ...
 $ rpcsq_rpq3     : num  2 4 2 3 5 0 6 4 0 0 ...
 $ rpcsq_rpq13    : num  4 16 19 2 16 4 15 23 12 16 ...
 $ rpcsq_cognitive: num  0 4 8 0 6 2 2 10 6 7 ...
 $ rpcsq_somatic  : num  6 14 13 3 15 2 12 10 6 6 ...
 $ rpcsq_emotional: num  0 2 0 2 0 0 7 7 0 3 ...

I'd like to produce ess x rpcsq scatterplots (5 scatterplots) with the points colored by group. I've tried the following:

> plot_scatterplot(tmp, by = "ess_score")

This works but obviously doesn't color the points. The following code however fails to produce the plots:

> plot_scatterplot(tmp, by = "ess_score", geom_point_args = list(col = "group"))
Error in grDevices::col2rgb(colour, TRUE) : invalid color name 'group'
> plot_scatterplot(tmp, by = "ess_score", geom_point_args = list(col = group))
Error in do.call("geom_point", geom_point_args) : 
  object 'group' not found
> plot_scatterplot(tmp, by = "ess_score", geom_point_args = list(group = "group", col = "group"))
Error in grDevices::col2rgb(colour, TRUE) : invalid color name 'group'

Does the functionality I'm looking for exist in the current iteration of DataExplorer? Thanks for any help you can be.

[boxuancui]

Thanks for using DataExplorer. For your need, you will have to tweak the source code a little. Copy & paste the following function and you should be able to pass group.

plot_scatterplot2 <- function(data, by, group, sampled_rows = nrow(data), geom_point_args = list(), title = NULL, ggtheme = theme_gray(), theme_config = list(), nrow = 3L, ncol = 3L, parallel = FALSE) {
  variable <- NULL
  if (!is.data.table(data)) data <- data.table(data)
  if (sampled_rows < nrow(data)) data <- data[sample.int(nrow(data), sampled_rows)]
  dt <- suppressWarnings(melt.data.table(data, id.vars = c(by, group), variable.factor = FALSE))
  feature_names <- unique(dt[["variable"]])
  layout <- DataExplorer:::.getPageLayout(nrow, ncol, length(feature_names))
  plot_list <- DataExplorer:::.lapply(
    parallel = parallel,
    X = layout,
    FUN = function(x) {
      ggplot(dt[variable %in% feature_names[x]], aes_string(x = by, y = "value", color = group)) +
        do.call("geom_point", geom_point_args) +
        coord_flip() +
        xlab(by)
    }
  )
  class(plot_list) <- c("multiple", class(plot_list))
  plotDataExplorer(
    plot_obj = plot_list,
    page_layout = layout,
    title = title,
    ggtheme = ggtheme,
    theme_config = theme_config,
    facet_wrap_args = list(
      "facet" = ~ variable,
      "nrow" = nrow,
      "ncol" = ncol,
      "scales" = "free_x",
      "shrink" = FALSE
    )
  )
}

Then just do:

plot_scatterplot2(tmp, by = "ess_score", group = "group")

I tested it on iris and it works fine:

plot_scatterplot2(iris, by = "Sepal.Length", group = "Species")

[boxuancui]

Please also keep this issue open. I might be able to add this in future versions, but I can't promise which one.

[araikes]

Thanks @boxuancui. I'll give it a shot.

[khughitt]

In general, having the ability to color (or assign other ggplot2 aesthetics) based on some groups defined in some column would be quite useful for many of the plotting functions (plot_density, etc.)