Advanced plotting

This tutorial will go over more advanced plotting functionality. Before reading this, you should take a look at the basic analysis and plotting tutorial. First, we’ll load in some example data. This dataset is an egg comprised of 30 subjects, who each performed 8 study/test blocks of 16 words each.

import quail
%matplotlib inline
egg = quail.load_example_data()

/usr/local/lib/python3.6/site-packages/pydub/utils.py:165: RuntimeWarning: Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work
  warn("Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work", RuntimeWarning)

Accuracy

accuracy = egg.analyze('accuracy')
accuracy.get_data().head()
0
Subject List
0 0 0.5000
1 0.5625
2 0.1250
3 0.5625
4 0.3125

By default, the analyze function will perform an analysis on each list separately, so when you plot the result, it will plot a separate bar for each list, averaged over all subjects:

ax = accuracy.plot()
../_images/advanced_plotting_5_0.png

We can plot the accuracy for each subject by setting plot_type='subject', and we can change the name of the subject grouping variable by setting the subjname kwarg:

ax = accuracy.plot(plot_type='subject', subjname='Subject Number')
../_images/advanced_plotting_7_0.png

Furthermore, we can add a title using the title kwarg, and change the y axis limits using ylim:

ax = accuracy.plot(plot_type='subject', subjname='Subject Number',
                title='Accuracy by Subject', ylim=[0,1])
../_images/advanced_plotting_9_0.png

In addition to bar plots, accuracy can be plotted as a violin or swarm plot by using the plot_style kwarg:

ax = accuracy.plot(plot_type='subject', subjname='Subject Number',
                title='Accuracy by Subject', ylim=[0,1], plot_style='violin')
ax = accuracy.plot(plot_type='subject', subjname='Subject Number',
                title='Accuracy by Subject', ylim=[0,1], plot_style='swarm')
../_images/advanced_plotting_11_0.png

We can also group the subjects. This is useful in cases where you might want to compare analysis results across multiple experiments. To do this we will reanalyze the data, averaging over lists within a subject, and then use the subjgroup kwarg to group the subjects into two sets:

accuracy = egg.analyze('accuracy', listgroup=['average']*8)
accuracy.get_data().head()
0
Subject List
0 average 0.367188
1 average 0.601562
2 average 0.742188
3 average 0.546875
4 average 0.867188 
ax = accuracy.plot(subjgroup=['Experiment 1']*15+['Experiment 2']*15)
../_images/advanced_plotting_14_0.png

Oops, what happened there? By default, the plot function looks to the List column of the df to group the data. To group according to subject group, we must tell the plot function to plot by subjgroup. This can be achieved by setting plot_type='subject':

ax = accuracy.plot(subjgroup=['Experiment 1']*15+['Experiment 2']*15, plot_type='subject')
../_images/advanced_plotting_16_0.png

If you also have a list grouping (such as first 4 lists / second 4 lists), you can plot the interaction by setting plot_type='split'. This will create a plot with respect to both the subjgroup and listgroup:

accuracy = egg.analyze('accuracy', listgroup=['First 4 Lists']*4+['Second 4 Lists']*4)
ax = accuracy.plot(subjgroup=['Experiment 1']*15+['Experiment 2']*15, plot_type='split')
../_images/advanced_plotting_18_0.png

Like above, these plots can also be violin or swarm plots:

ax = accuracy.plot(subjgroup=['Experiment 1']*15+['Experiment 2']*15, plot_type='split', plot_style='violin')
ax = accuracy.plot(subjgroup=['Experiment 1']*15+['Experiment 2']*15, plot_type='split', plot_style='swarm')
../_images/advanced_plotting_20_0.png

Memory fingerprints

The Memory Fingerprint plotting works exactly the same as the the accuracy plots, with the except that plot_type='split' only works for the accuracy plots, and the default plot_style is a violinplot, instead of a barplot.

fingerprint = egg.analyze('fingerprint', listgroup=['First 4 Lists']*4+['Second 4 Lists']*4)
ax = fingerprint.plot(subjgroup=['Experiment 1']*15+['Experiment 2']*15, plot_type='subject')
ax = fingerprint.plot(subjgroup=['Experiment 1']*15+['Experiment 2']*15, plot_type='list')
../_images/advanced_plotting_23_0.png

Other analyses

Like the plots above, spc, pfr and lagcrp plots can all be plotted according to listgroup or subjgroup by setting the plot_type kwarg.

Plot by list grouping

listgroup = ['First 4 Lists']*4+['Second 4 Lists']*4
plot_type = 'list'

spc = egg.analyze('spc', listgroup=listgroup)
ax = spc.plot(plot_type=plot_type, ylim=[0, 1])

pfr = egg.analyze('pfr', listgroup=listgroup)
ax = pfr.plot(plot_type=plot_type)

lagcrp = egg.analyze('lagcrp', listgroup=listgroup)
ax = lagcrp.plot(plot_type=plot_type)

/Users/andrewheusser/Documents/github/quail_contextlab/quail/analysis/lagcrp.py:129: RuntimeWarning: Mean of empty slice
  return np.nanmean(lagcrp, axis=0)
../_images/advanced_plotting_26_1.png

Plot by subject grouping

listgroup=['average']*8
subjgroup = ['Experiment 1']*15+['Experiment 2']*15
plot_type = 'subject'

spc = egg.analyze('spc', listgroup=listgroup)
ax = spc.plot(subjgroup=subjgroup, plot_type=plot_type, ylim=[0,1])

pfr = egg.analyze('pfr', listgroup=listgroup)
ax = pfr.plot(subjgroup=subjgroup, plot_type=plot_type)

lagcrp = egg.analyze('lagcrp', listgroup=listgroup)
ax = lagcrp.plot(subjgroup=subjgroup, plot_type=plot_type)

/Users/andrewheusser/Documents/github/quail_contextlab/quail/analysis/lagcrp.py:129: RuntimeWarning: Mean of empty slice
  return np.nanmean(lagcrp, axis=0)
../_images/advanced_plotting_28_1.png