timelyportfolio/sunburstR

sumNodes=FALSE makes sunburst not work

madlogos opened this issue · 2 comments

When I run the following example code, the plot fails to render.

sequences <- read.csv(
     system.file("examples/visit-sequences.csv",package="sunburstR")
     , header = FALSE
     , stringsAsFactors = FALSE
)[1:100, ]
sunburst(sequences, sumNodes=FALSE)

I am using sunburstR 2.0.0 installed from CRAN. Could you look into this?

To be clear, you modified the existing example to use sumNodes=FALSE?

The sumNodes argument is a recent addition that was added to deal with a specific data format/structure. The data format/structure used in that example is not appropriate for using the sumNodes=FALSE option.

An example that uses the sumNodes=FALSE option is here:
https://github.com/timelyportfolio/sunburstR/blob/master/inst/examples/example_treemap.R

There's a short explanation of why it was added here:
https://github.com/timelyportfolio/sunburstR/blob/1819c49b288a3500f52a6b98f6d0242ce0b77837/vignettes/sunburst-2-0-0.Rmd#sum-nodes

and the issue for adding it (with a much more detailed explanation of its purpose is here:
#62


There could be something said for adding a bit of checking to throw an error before displaying an empty plot, as happens with your modified example.

Thank you, @cjyetman. I modified the test code according to the example you provided. Now it worked correctly.

Vectorize(function(pkg) invisible(require(pkg, character.only=TRUE, quiet=TRUE)))(
    c("dplyr", "sunburstR", "treemap", "d3r")
)

sequences <- read.csv(
    system.file("examples/visit-sequences.csv",  package="sunburstR")
    , header = FALSE
    , stringsAsFactors = FALSE
)[1:100, ]

seqdat <- dplyr::bind_rows(lapply(
    strsplit(as.character(sequences[[1]]), "-"), function(rw) 
        data.frame(t(rw), stringsAsFactors = FALSE)))
seqdat$size <- sequences$V2

tm <- treemap(seqdat, index=paste("X", 1:6, sep=""), vSize="size", draw=FALSE)
tmnest <- d3_nest(tm$tm[, c(paste("X", 3:6, sep=""), "vSize", "color")],
                  value_cols = c("vSize", "color"))
sunburst(
    data = tmnest,
    valueField = "vSize",
    count = TRUE,
    sumNodes = FALSE
)

I agree that it will be super if there is some data structure validation and error message before rendering the plot. Now I see that sumNodes is happier to work with treemap lists.