grammar of graphics - ideas and how to in R / Matlab

make-gg-plots.R

# re-create plot for vision group talk 20170309
#
# ds
library(ggplot2)
library(dplyr)
library(tidyr)
# if you get an error about unknown packages, you will probably have to run the following first:
# install.packages(c("ggplot2","tidyr","dplyr"))

# random seed - probably not strictly necessary
set.seed(12345)

# data for figure 3b from Zamboni et al (2016)
d = read.csv('xy-dip-data+.csv')

d$subject = factor(d$subject) # make it categorical

# ceate ggplot object with the appropriate data -> aes mappings
q <- ggplot(data = d, 
            aes(x = absent, y = present)  )

# decide on the ``geom``- this will also display it
q + geom_point()

# actually, we want the full mapping
p <- ggplot(data = d, 
            aes(x = absent, y = present, size = coherence, color = subject)  )

p + geom_point()

## now deal with ``scale`` + the annotations.

finalPlot <- p + geom_point() +
  scale_size( breaks = c(4,7,13,25) ) +  # could also do breaks = unique(d$coherence)
  scale_x_continuous( limits = c(0, 0.1) ) + 
  scale_y_continuous( limits = c(0, 0.1) ) +
  coord_fixed() + 
  geom_abline(slope = 1, intercept = 0, size = 1, color="Gray")

# and display
finalPlot + theme_bw()

# Themes!
# can easily change the look of your plots (this is also customisable)

p1 <- finalPlot + theme_classic()
p2 <- finalPlot + theme_minimal()
p3 <- finalPlot + theme_dark()
p4 <- finalPlot + theme_light()

source("http://peterhaschke.com/Code/multiplot.R") 
multiplot(p1, p2, p3, p4, cols = 2)

# Facets!
finalPlot + 
  facet_wrap(~subject) +
  theme_minimal() + 
  labs(title = "Data for different subjects")

# Facets!
finalPlot + 
  facet_wrap(~coherence) +
  theme_minimal() + 
  labs(title = "Data by coherence")


# ... but how if we want to do this with bar plots?
# data need to be in tidy format...
e = gather(d, condition, statistic, -X, -subject, -coherence)
head(e)
r <- ggplot(data = e, 
            aes(x = subject, y = statistic, 
                fill = condition, color = condition)  )
r + geom_bar(stat= "identity", position="dodge")  

UsingGramm.m

%% introduction to gramm
%
% ds 20170309
%
% the following is adapted from the website for GRAMM
% https://github.com/piermorel/gramm

clear all
close all

load carbig.mat %Load example dataset about cars
origin_region=num2cell(org,2); %Convert origin data to a cellstr

%% GRAMM uses object-oriented approach
%  CREATE a gramm object that contains the mapping from data -> AES
figure
g=gramm('x',Model_Year,'y',MPG,'color',Cylinders, 'subset',Cylinders~=3 & Cylinders~=5)
% Plot raw data as points
g.geom_point()
% Set appropriate names for legends
g.set_names('column','Origin','x','Year','y','Fuel economy (MPG)','color','# Cylinders')
%Set figure title
g.set_title('Fuel economy of new cars between 1970 and 1982')
% Do the actual drawing
g.draw()



%% try again, but add a facet grid and linear fit
figure
h=gramm('x',Model_Year,'y',MPG,'color',Cylinders, 'subset',Cylinders~=3 & Cylinders~=5)
h.geom_point()
% Plot linear fits of the data with associated confidence intervals
h.stat_glm()
% Subdivide the data in subplots horizontally by region of origin
h.facet_grid([],origin_region)
% and draw this one
h.draw()

xy-dip-data+.csv