plot_prop2: Plot the Proportion of Each Combination of Two Predictors'...

In AndrewKostandy/MLtoolkit: Functions to Help with Machine Learning & Feature Engineering Tasks

Description

Plot the proportion of each combination of two predictors' levels having a specific outcome level and the 95% confidence interval of the proportion. Useful for assessing the presence/effects of an interaction between categorical predictors on an outcome.

Usage

  plot_prop2(data, predictor1, predictor2, outcome, ref_level,
             ref_value = 0.5, add_n = FALSE, width = 0, flip = FALSE)

Arguments

data	The dataframe containing the predictor and outcome variables.
predictor1	The first predictor variable name. This will be shown on the x-axis.
predictor2	The second predictor variable name. This will be identified by color and shape in the legend.
outcome	The outcome variable name.
ref_level	A string providing the specific outcome level to show the proportions for.
ref_value	Defaults to 0.5. A numeric providing the proportion value of interest to compare to.
add_n	Defaults to TRUE. A logical indicating whether to add the total number of observations of each level to the plot.
width	Defaults to 0. A numeric providing the width of the error bars.
flip	Defaults to FALSE. A logical indicating whether to flip the plot.

Value

Returns a ggplot object with the first predictor levels on the x-axis and the proportion of values in the ref_level of the outcome on the y-axis. The second predictor variable is identified by shape and color in the legend.

Figures

Author(s)

Andrew Kostandy ([e-mail])

Examples

Titanic2 <- as.data.frame(Titanic)
Titanic2 <- Titanic2[rep(seq(1:nrow(Titanic2)), Titanic2$Freq), -5]

plot_prop2(Titanic2, Class, Sex, Survived, "Yes", ref_value = 0.323,
           add_n = FALSE, width = 0, flip = FALSE)

AndrewKostandy/MLtoolkit documentation built on May 7, 2019, 9:51 p.m.