Repository files navigation

An R package to ease data visualization.

The aim of this package is to make visualization an early part of the data analysis process by automating a few common plotting tasks.

In terms of design, it has three general principles:

• Flexibility: splot prefers to try to make a reasonable plot from any input, rather than erroring out.
• Minimal specification: You should be able to make most plots with just a formula input, and the same formula should generally be compatible with multiple plot types, regardless of variable types.
• Tweakability: Though splot is focused on quick, automated plotting, you should be able to adjust any aspect of the display with additional arguments, if you find a plot you want to display elsewhere.

By entering a formula as the first argument in thesplot function (e.g.,splot(y ~ x)), you can make

• Density distributions (overlaid on histograms when there is noby variable)
• Scatter plots with prediction lines
• Bar or line graphs with error bars

For each type, multipley variables or data at levels of aby variable are shown in the same plot frame,
and data at levels of one or twobetween variables are shown in separate plot frames, organized in a grid.

• Introduction
• Documentation
• Style Guide
• Gallery
• Applied examples:Exploring Data |Refining a Result

Download R fromr-project.org.

Release (version 0.5.4)

install.packages("splot")

Development (version 0.5.5)

# install.packages("remotes")remotes::install_github("miserman/splot")

Then load the package:

library(splot)

Make some data: random group and x variables, and a y variable related to x:

group= rep(c("group 1","group 2"),50)x= rnorm(100)y=x*.5+ rnorm(100)

The distribution of y:

splot(y)

A scatter plot between y and x:

splot(y~x)

Same data with a quadratic model:

splot(y~x+x^2+x^3)

Same data separated by group:

splot(y~x*group)

Could also separate by median or standard deviations of x:

splot(y~x*x)splot(y~x*x,split="sd")

Summarize with a bar plot:

splot(y~x*group,type="bar")

Two-level y variable with a probability prediction line:

# make some new data for this example:# a discrete y variable and related x variable:y_bin= rep(c(1,5),50)x_con=y_bin*.4+ rnorm(100)# lines = "prob" for a prediction line from a logistic model:splot(y_bin~x_con,lines="prob")