demogif

Better {shiny} Support

It now integrates better into shiny apps. There is a new function parcats_demo() which let’s you interactively explore all the different parameters of easyalluvial::alluvial_wide() and parcats::parcats(). You can see how the alluvial plot and the derived interactive parcats widget look like with different parameters.

demo

Update {plotly.js}

In order to be 100% compatible with R plotly. plotly.js that is shipped with parcats has been upgraded to v2.5.1

Parcats from Alluvial Plot

suppressPackageStartupMessages(require(tidyverse))
suppressPackageStartupMessages(require(easyalluvial))
suppressPackageStartupMessages(require(parcats))
suppressPackageStartupMessages(require(parsnip))
p = alluvial_wide(mtcars2, max_variables = 5)

parcats(p, marginal_histograms = TRUE, data_input = mtcars2)

Partial Dependence Alluvial Plots

Machine Learning models operate in a multidimensional space and their response is hard to visualise. Model response and partial dependency plots attempt to visualise ML models in a two dimensional space. Using alluvial plots or parallel categories diagrams we can increase the number of dimensions.

• Visualise model response with alluvial plots

df <- select(mtcars2, -ids)

m <- parsnip::rand_forest(mode = "regression") %>%
  parsnip::set_engine("randomForest") %>%
  parsnip::fit(disp ~ ., df)

p <- alluvial_model_response_parsnip(m, df, degree = 4, method = "pdp")
## Getting partial dependence plot preditions. This can take a while. See easyalluvial::get_pdp_predictions() `Details` on how to use multiprocessing
parcats(p, marginal_histograms = TRUE, imp = TRUE, data_input = df)