What's New in tidyvpc? • tidyvpc

tidyvpc 1.4.0

Additional Binning Methods

The following additional binning methods from classInt have been made available in tidyvpc.

see ?classIntervals ‘style’ descriptions for applicable arguments for each selected binning method.

library(tidyvpc)

## tidyvpc is part of Certara.R!
## Follow the link below to learn more about PMx R package development at Certara.
## https://certara.github.io/R-Certara/

obs_data <- obs_data[MDV == 0]
sim_data <- sim_data[MDV == 0]

`headtails`

observed(obs_data, x = TIME, y = DV) |>
  simulated(sim_data, y = DV) |>
  binning(bin = "headtails") |>
  plot()

Including additional thr argument.

observed(obs_data, x = TIME, y = DV) |>
  simulated(sim_data, y = DV) |>
  binning(bin = "headtails", thr = 0.55) |>
  plot()

`maximum`

observed(obs_data, x = TIME, y = DV) |>
  simulated(sim_data, y = DV) |>
  binning(bin = "maximum") |>
  plot()

## Warning in RColorBrewer::brewer.pal(n, pal): n too large, allowed maximum for palette Set1 is 9
## Returning the palette you asked for with that many colors

## Warning: Removed 2 rows containing missing values (geom_point).

Including additional nbins argument.

observed(obs_data, x = TIME, y = DV) |>
  simulated(sim_data, y = DV) |>
  binning(bin = "maximum", nbins = 7) |>
  plot()

`box`

observed(obs_data, x = TIME, y = DV) |>
  simulated(sim_data, y = DV) |>
  binning(bin = "box") |>
  plot()

Including additional iqr_mult and type argument.

observed(obs_data, x = TIME, y = DV) |>
  simulated(sim_data, y = DV) |>
  binning(bin = "box", iqr_mult = 4) |>
  plot()

# additional (quantile) type arg
observed(obs_data, x = TIME, y = DV) |>
  simulated(sim_data, y = DV) |>
  binning(bin = "box", type = 3) |>
  plot()

Additional flexibility for binless() + predcorrect()

Users may now execute predcorrect() either before, or after calling binless(loess.ypc=TRUE). Previously, you were required to execute predcorrect() before binless(loess.ypc=TRUE), otherwise you’d receive an error.

The following code below produces equivalent output:

observed(obs_data, x = TIME, y = DV ) |>
  simulated(sim_data, y = DV) |>
  stratify(~ GENDER) |>
  predcorrect(pred=PRED) |> #before binless()
  binless(loess.ypc=TRUE) |>
  vpcstats() |>
  plot()

observed(obs_data, x = TIME, y = DV ) |>
  simulated(sim_data, y = DV) |>
  stratify(~ GENDER) |>
  binless(loess.ypc=TRUE) |> 
  predcorrect(pred=PRED) |> #after binless()
  vpcstats() |>
  plot()

tidyvpc 1.3.0

An overview of updates to plot() function in tidyvpc v1.3.0

Set plot output dimensions:

knitr::opts_chunk$set(fig.width=12, fig.height=8, dpi = 300)

library(tidyvpc)
library(magrittr)
obs_data <- obs_data[MDV == 0]
sim_data <- sim_data[MDV == 0]

vpc <- observed(obs_data, x=TIME, y=DV) %>%
  simulated(sim_data, y=DV) %>%
  stratify(~ GENDER) %>%
  binless() %>%
  vpcstats()

plot(vpc)

vpc <- observed(obs_data, x=TIME, y=DV) %>%
  simulated(sim_data, y=DV) %>%
  stratify(~ GENDER + STUDY) %>%
  binning(bin = "jenks", nbins = 8) %>%
  vpcstats()

plot(vpc)

vpc <- observed(obs_data, x=TIME, y=DV) %>%
  simulated(sim_data, y=DV) %>%
  stratify(GENDER ~ STUDY) %>%
  binning(bin = "kmeans", nbins = 6) %>%
  vpcstats()

plot(vpc)

We can use facet = TRUE argument to facet continuous VPC by quantile or facet categorical VPC by predicted probability.

vpc <- observed(obs_data, x=TIME, y=DV) %>%
  simulated(sim_data, y=DV) %>%
  binless() %>%
  vpcstats()

plot(vpc, facet = TRUE, point.alpha = 0.1, point.size = 1, ribbon.alpha = 0.2)

vpc <- observed(obs_cat_data, x = agemonths, yobs = zlencat) %>%
  simulated(sim_cat_data, ysim = DV) %>%
  binless() %>% 
  vpcstats(vpc.type = "categorical")

plot(vpc, facet = TRUE, legend.position = "bottom")

Changing point size, point alpha, point shape, point stroke, and ribbon alpha

Setup categorical VPC.

vpc <- observed(obs_cat_data, x = agemonths, yobs = zlencat) %>%
  simulated(sim_cat_data, ysim = DV) %>%
  binning(bin = round(agemonths, 0)) %>%
  vpcstats(vpc.type = "categorical")

Adjust point size.

plot(vpc, point.size = 4)

Setup continuous VPC.

vpc <- observed(obs_data, x=TIME, y=DV) %>%
  simulated(sim_data, y=DV) %>%
  binless() %>%
  vpcstats()

plot(vpc, point.size = 1.5, point.stroke = 2.5, point.alpha = 0.1, ribbon.alpha = 0.05)

plot(vpc, point.size = 5, point.stroke = 0.3, point.shape = "triangle")

plot(vpc, point.size = 7, point.shape = "square-fill", point.alpha = 0.1, ribbon.alpha = 0.5)

What’s New in tidyvpc?