CODEWITHSUNDEEP
rplotlyr-plotly
C. Braun
C. Braun

Reputation: 5211

Filter a plotly line chart based on categorical variable

I would like to filter a chart created with plotly, based on a column of discrete values in my data. The end goal is to be able to use buttons to update the filter value, so I do not want to filter the data beforehand.

library(plotly)    

df <- data.frame(group1 = rep(c('low', 'high'), each = 25),
                 x = rep(1:5, each = 5),
                 group2 = letters[1:5],
                 y = c(runif(25, 0, 2), runif(25, 3, 5)))

plot_ly(df, x = ~x, y = ~y, type = 'scatter',
        mode = 'line',
        color = ~group2,
        transforms = list(
            list(
                type = 'filter',
                target = ~group1,
                operation = '=',
                value = 'high'
            )
        )
)

I expected this to give the following chart:

enter image description here

but instead it gives this:

enter image description here

It seems to be filtering on the wrong variable. Why isn't the data being filtered the way I expect?

Upvotes: 3

Views: 6190

Answers (1)

LmW.
LmW.

Reputation: 1364

It seems the issue is that the target argument of the transforms list can only take the name of a plot_ly attribute, not raw data. This code works:

library(plotly)    

set.seed(1)
df <- data.frame(group1 = rep(c("low", "high"), each = 25),
                 x = rep(1:5, each = 5),
                 group2 = letters[1:5],
                 y = c(runif(25, 0, 2), runif(25, 3, 5)))

plot_ly(df, x = ~x, y = ~y, customdata=~group1, type = 'scatter',
        mode = 'line',
        color = ~group2,
        transforms = list(
          list(
            type = 'filter',
            target = 'customdata',
            operation = '=',
            value = 'high'
          )
        )
)

and produces the same chart as this 

plot_ly(df %>% filter(group1=="high"), x = ~x, y = ~y, type = 'scatter',
        mode = 'line',
        color = ~group2
)

Of course, instead of using customdata, you can replace it with ids, etc, an attribute that plot_ly allows but doesn't affect the aesthetics of your chart.

Upvotes: 9

Related Questions