metaplot (version 0.2.7)

metaplot.data.frame: Create Metaplot for Data Frame.

Description

Metaplot creates univariate, bivariate, or multivariate plots depending on the number and types of variables represented by the anonymous arguments. Types are either numeric (NUM, e.g. real, integer) or categorical (CAT, e.g. factor, character). A variable stored as numeric that nonetheless has an encoded guide attribute will be treated as categorical.

Usage

# S3 method for data.frame
metaplot(x, ...,
  univariate = getOption("metaplot_univariate", "dens"),
  mixedvariate = getOption("metaplot_mixedvariate", "boxplot"),
  bivariate = getOption("metaplot_bivariate", "scatter"),
  multivariate = getOption("metaplot_multivariate", "corsplom"),
  categorical = getOption("metaplot_categorical", "categorical"))

Arguments

x

object

...

passed arguments

univariate

function for univariate arguments

mixedvariate

function for bivariate combinations of numeric and categoral arguments

bivariate

function for arguments that resolve to two numerics (see rules)

multivariate

function for more than two numeric arguments

categorical

function for categorical arguments

Details

Design your plot by specifying y variables (optional), the x variable, the groups variable (optional) and the conditioning variables (i.e., facets, optional).

The single groups variable, if any, is the first categorical in the third position or later. An earlier categorical gives a bivariate plot, e.g. horizontal boxplot (first position) or vertical boxplot (second postion) causing groups and facets to be ignored.

The x variable is the last variable before groups, if present.

The y variables are those before x. If none, the result is univariate. If one, the result is typically a boxplot or scatterplot, depending on x. Several numeric y followed by a numeric x are treated as multivariate (scatterplot matrix). But if all y have the same guide attribute and it is different from that for x, the result is bivariate (i.e, an overlay scatterplot).

The groups argument is only relevant for bivariate plots with one y. For multiple y (overlay), the sources of y are the implied groups: any trailing categorical arguments are treated as facets. To specify facets without specifying groups, let groups be empty, e.g., (metaplot(y, x, , facet1, facet2)).

Template designs follow; substitute behaviors by setting global options (see argument list).

• NUM: univariate (densityplot)

• CAT: categorical (unimplemented)

• NUM, CAT:mixedvariate (vertical boxplot)

• CAT, NUM:mixedvariate (horizontal boxplot)

• NUM, NUM:bivariate (scatterplot)

• CAT, CAT: categorical (unimplemented)

• NUM, NUM, CAT: grouped bivariate (grouped scatterplot)

• NUM, NUM,, CAT: non-grouped bivariate with one facet

• NUM, NUM, CAT, CAT: grouped bivariate with one facet

• NUM, NUM, CAT, CAT, CAT: grouped bivariate with two facets

• NUM, NUM, NUM: multivariate, or grouped bivariate for overlay

• NUM, NUM, NUM, CAT multivariate, or faceted bivariate for overlay

• NUM, NUM, NUM, CAT, CAT multivariate, or bivariate with two facets for overlay

See Also

Other metaplot: categorical, metaplot.folded, metaplot

Other univariate plots: dens.data.frame, dens.folded, dens_data_frame, metaplot.folded

Other bivariate plots: boxplot.folded, metaplot.folded, scatter.data.frame, scatter.folded, scatter_data_frame

Other multivariate plots: corsplom.data.frame, corsplom.folded, corsplom_data_frame, metaplot.folded

Examples

# NOT RUN {
library(magrittr)
library(dplyr)
library(csv)
library(nlme)
x <- Theoph

# mixed effects model
m1 <- nlme(
  conc ~ SSfol(Dose, Time, lKe, lKa, lCl),
  data = x,
  fixed = lKe + lKa + lCl ~ 1,
  random = lKe + lKa + lCl ~ 1
)

# some numeric and categorical properties
x %<>% mutate(arm = ifelse(as.numeric(as.character(Subject)) %% 2 == 0, 1, 2))
x %<>% mutate(site = ifelse(as.numeric(as.character(Subject)) < 7, 1, 2))
x %<>% mutate(pred = predict(m1,level = 0))
x %<>% mutate(ipred = predict(m1))
x %<>% mutate(res = residuals(m1))
x %<>% mutate(sres = residuals(m1, type = 'pearson'))
r <- ranef(m1)
r$Subject <- rownames(r)
x %<>% left_join(r)
# metadata
attr(x$Subject,'label') <- 'subject identifier'
attr(x$Wt,'label') <- 'subject weight'
attr(x$Dose,'label') <- 'theophylline dose'
attr(x$Time,'label') <- 'time since dose administration'
attr(x$conc,'label') <- 'theophylline concentration'
attr(x$arm,'label') <- 'trial arm'
attr(x$site,'label') <- 'investigational site'
attr(x$pred,'label') <- 'population-predicted concentration'
attr(x$ipred,'label') <- 'individual-predicted conentration'
attr(x$res,'label') <- 'residuals'
attr(x$sres,'label') <- 'standardized residuals'
attr(x$lKe,'label') <- 'natural log of elimination rate constant'
attr(x$lKa,'label') <- 'natural log of absorption rate constant'
attr(x$lCl,'label') <- 'natural log of clearance'
attr(x$Subject,'guide') <- '....'
attr(x$Wt,'guide') <- 'kg'
attr(x$Dose,'guide') <- 'mg/kg'
attr(x$Time,'guide') <- 'h'
attr(x$conc,'guide') <- 'mg/L'
attr(x$arm,'guide') <- '//1/Arm A//2/Arm B//'
attr(x$site,'guide') <- '//1/Site 1//2/Site 2//'
attr(x$pred,'guide') <- 'mg/L'
attr(x$ipred,'guide') <- 'mg/L'
x %>% unpack %>% as.csv('theoph.csv')
# }
# NOT RUN {
library(magrittr)
library(dplyr)
library(csv)
x <- as.csv(system.file(package = 'metaplot', 'extdata/theoph.csv'))
x %<>% pack
# sample plots
x %>% metaplot(conc)
#x %>% metaplot(site)
x %>% metaplot(Wt, arm)
x %>% metaplot(arm, Wt)
x %>% metaplot(conc, Time)
#x %>% metaplot(arm, site)
x %>% metaplot(conc, Time, Subject)
x %>% metaplot(conc, Time, , Subject)
x %>% metaplot(conc, Time, Subject, site)
x %>% metaplot(conc, Time, Subject, site, arm)
x %>% metaplot(lKe, lKa, lCl)
x %>% metaplot(conc, ipred, Time)
x %>% metaplot(conc, ipred, Time, Subject)
x %>% metaplot(conc, ipred, Time, site, arm)
x %>% metaplot(res, conc, yref = 0, ysmooth = T, conf = T, grid = T, loc = 1)
x %>% metaplot(conc, Time, panel = panel.smoothScatter)

# }
# NOT RUN {
# }