def group():
'''## Grouping
One of the more interesting features of this plot is the ability to group a
set of streams within a stream. the inner streamgroup can be aligned within the
bounds of the stream in which it is contained. If the values of a stream
that contains a streamgroup are set to `None`, the stream size will be inferred
from the sum of its inner streams. If the `normalize` flag is used, the inner
streams will be proportionately scaled to fit within the bounds of the outer
stream.
The labels can be moved to a legend on the right by using `legend=True` or a
float between 0.0 and 1.0 to indicate the relative size of the legend.'''
import matplotlib.pyplot as plt
import streamgraph as sg
from matplotlib import cm
substreams = [
sg.Stream([4.0, 2.5, 2.0, 1.0], label='sub1'),
sg.Stream([2.0, 1.5, 2.0, 2.0], label='sub2'),
sg.Stream([ 3.0, 0.5, 3.0], label='sub3', offset=1),
]
substreamset = sg.StreamSet(substreams, normalize=True,
facecolors=cm.gist_gray)
main_streams = [
sg.Stream([2.0, 1.5, 2.0, 2.0], label='main1', streamset=substreamset),
sg.Stream([4.0, 2.5, 2.0, 1.0], label='main2'),
sg.Stream([ 3.0, 0.5, 3.0], label='main3', offset=1),
]
streamset = sg.StreamSet(main_streams,
pad=0.2, margin=0.5)
fig = plt.figure(figsize=(8, 3))
sg.streamgraph(plt, streamset, legend=0.25)
def skills():
'''## Finally
And finally, I get to use this for the purpose I intended: a graphic to add to my CV
to concisely communicate my skill history in a memorable way for anyone that
might find me useful.'''
import matplotlib.pyplot as plt
import streamgraph as sg
from matplotlib import cm
import pandas
import numpy as np
io = open('examples/my_skills.csv')
df = pandas.io.parsers.read_csv(io, index_col=0, header=0)
streamsets = {} #python, js, etc
for h in df.columns:
main_h, sub_h = h.split('/')
streamsets.setdefault(main_h,{})[sub_h] = sg.Stream(df[h], label=sub_h)
main_streams = []
for k in streamsets:
streams = streamsets[k].values()
o = min(s.offset for s in streams)
lab_stream = sg.Stream([1], sort_values=[99], offset=o, label=k, alpha=0.0)
ss = sg.StreamSet(streams+[lab_stream], pad=0.25, facecolors=cm.Greys)
s = sg.Stream(None, label='', streamset=ss)
main_streams.append(s)
main_streamset = sg.StreamSet(main_streams, facecolors=cm.jet,
normalize=False, pad=.25)
fig = plt.figure(figsize=(18, 5))
sg.streamgraph(plt, main_streamset, xs=df.index, legend=False)
plt.tight_layout()
def formatted():
'''## Formatting
Streams can have a label and custom color and shape formatting.
Colormaps can be used to automtacially color all the streams
in a streamset or each stream can be individually colored.
The transitions between values in a stream can be made smoother or sharper
by tweaking the control points.
Padding can be added between streams and a margin can be added to the
streamset.
The streamset can be aligned within it's bounds. In the bounds of
the plot axis in can be centered by using `align=0.5`.
In the following example
we have harshened the transitions, brashened the colors, and aligned the streams
with the bottom of the plot with `align=0.0`.'''
import matplotlib.pyplot as plt
import streamgraph as sg
from matplotlib import cm
streams = [
sg.Stream([4.0, 2.5, 2.0, ], label='gluon'),
sg.Stream([2.0, 1.5, 2.0, 2.0,], label='muon'),
sg.Stream([ 3.0, 0.5, 3.0,], label='quark', offset=1,
facecolor='#ff69b4', edgecolor='b'),
]
streamset = sg.StreamSet(streams, pad=0.5, margin=1.0, align=0.0,
facecolors=cm.jet, edgecolors='r')
fig = plt.figure(figsize=(8, 3))
sg.streamgraph(plt, streamset, transsize=0.2, transshape=0.1,
xs='up down charm strange'.split())
ax = fig.axes[0]
plt.subplots_adjust(bottom=0.2)
plt.setp(ax.xaxis.get_minorticklabels(), rotation=45)
def gdp_indv():
'''## GDP Individual
Let's try this out on real-world data: country GDPs. You'll need pandas for
this example. China's recent growth is immediately apparant. It is also obvious
that Brazil has been doing well. Relative GDP are also easily visible, for
example, Russia and Australia have similar GDP.'''
import matplotlib.pyplot as plt
import streamgraph as sg
from matplotlib import cm
import pandas
from matplotlib.ticker import FuncFormatter
def trillions(x, pos):
return '$%1.1fT' % (x*1e-12)
formatter = FuncFormatter(trillions)
io = open('examples/gdp.csv')
df1 = pandas.io.parsers.read_csv(io, index_col=0)
df2 = df1.pivot('Year', 'Country Code')
df3 = df2['Value']
df4 = df3.ix[2001:]
codes = ['USA','JPN','CHN', 'DEU', 'FRA', 'BRA', 'GBR', 'ITA', 'RUS', 'IND',
'CAN', 'AUS']
streams = [sg.Stream(df4[k], label=k) for k in codes]
streamset = sg.StreamSet(streams, pad=1.0e12, facecolors=cm.Accent)
fig = plt.figure(figsize=(10, 4))
sg.streamgraph(plt, streamset, xs=df4.index, legend=True)
ax = fig.axes[0]
ax.yaxis.set_major_formatter(formatter)
def braid():
'''## Recursion
A streamset can be associated with one of it's own inner streams, setting up
the potential for infinite recursion. The `RECURSION_LIMIT` value prevents
this. I'm not sure why anyone would want this except that it does make a
pretty fractal.'''
import matplotlib.pyplot as plt
import streamgraph as sg
import numpy as np
sg.RECURSION_LIMIT = 4
patt = '''
123
213
231
321
312
132
'''
l = [map(float,list(e.strip())) for e in patt.strip().split('\n')]
a = np.argsort(np.array(l).T,0)+1
aa = np.hstack((a,a))
streams = [sg.Stream(aa[i], offset=0) for i in range(3)]
streamset = sg.StreamSet(streams, pad=0.0, normalize=True)
streams[1].streamset = streamset
fig = plt.figure(figsize=(14, 2))
sg.streamgraph(plt, streamset)
def simple2():
'''## Alternate definition
an equivalent definition of stream offset using `None`'''
import matplotlib.pyplot as plt
import streamgraph as sg
streams = [
sg.Stream([ 4.0, 2.5, 2.0, ]),
sg.Stream([ 2.0, 1.5, 2.0, 2.0,]),
sg.Stream([None, 3.0, 0.5, 3.0,]),
]
streamset = sg.StreamSet(streams)
sg.streamgraph(plt, streamset)
def simple1():
'''## Basic Usage
A stream is ultimately just a sequence of values.
and a streamset is a collection of streams.
Streams have start points and end points defined by their offset and number
of values in the stream. Streams are useful when you want to visualize
how a quantity varies (usually over time) as well as it's ranking in a set
of other quantities.'''
import matplotlib.pyplot as plt
import streamgraph as sg
streams = [
sg.Stream([4.0, 2.5, 2.0, ]),
sg.Stream([2.0, 1.5, 2.0, 2.0,]),
sg.Stream([ 3.0, 0.5, 3.0,], offset=1),
]
streamset = sg.StreamSet(streams)
sg.streamgraph(plt, streamset)
def gdp_group():
'''## GDP Grouped
More of the story can be told when we group some of the countries
into larger economic entities. In this case we create a streamgroup for the
EU and for North America and it is interesting to note the EU's success as an
entity.'''
import matplotlib.pyplot as plt
import streamgraph as sg
from matplotlib import cm
import pandas
from matplotlib.ticker import FuncFormatter
def trillions(x, pos):
return '$%1.1fT' % (x*1e-12)
formatter = FuncFormatter(trillions)
io = open('examples/gdp.csv')
df1 = pandas.io.parsers.read_csv(io, index_col=0)
df2 = df1.pivot('Year', 'Country Code')
df3 = df2['Value']
df4 = df3.ix[2001:]
codes = ['NAC', 'EUU', 'JPN', 'CHN', 'BRA', 'RUS', 'IND', 'AUS',]
streams = [sg.Stream(df4[k], label=k) for k in codes]
eu_stream = streams[1]
codes = ['DEU', 'FRA', 'GBR', 'ITA', ]
substreams = [sg.Stream(df4[k], label=k) for k in codes]
eu_stream.streamset = sg.StreamSet(substreams, pad=0, align=0)
eu_stream.facecolor = '.8'
na_stream = streams[0]
codes = ['USA', 'CAN',]
substreams = [sg.Stream(df4[k], label=k) for k in codes]
na_stream.streamset = sg.StreamSet(substreams, pad=0, align=0)
mainstreamset = sg.StreamSet(streams, pad=1.0e12, facecolors=cm.Accent)
fig = plt.figure(figsize=(10, 4))
sg.streamgraph(plt, mainstreamset, xs=df4.index, legend=True)
ax = fig.axes[0]
ax.yaxis.set_major_formatter(formatter)
# sort streamgraphs appropriately
argsort_onset = streamgraph.argsort_onset(streams)
streams = streams[argsort_onset]
matrix = matrix[argsort_onset]
colors = colors[argsort_onset]
# argsort_inside_out = streamgraph.argsort_inside_out(streams)
# streams = streams[argsort_inside_out]
# colors = colors[argsort_inside_out]
# filter out some clones
if options.filter == 'none':
filter_idxs = np.ones(streams.shape[0]) > 0 # all streams
elif options.filter == 'seen2':
filter_idxs = np.sum(streams > 0, axis=1) >= 2 # seen twice
elif options.filter == 'sum2':
filter_idxs = np.sum(matrix, axis=1) >= 2 # sum=2
elif options.filter == 'sum3':
filter_idxs = np.sum(matrix, axis=1) >= 3 # sum=3
else:
raise ValueError, "what filter do you want me to use?"
fig = plt.figure(figsize=(24,16))
ax = fig.add_subplot(111)
streamgraph.streamgraph(ax, streams[filter_idxs], x=times, colors=colors[filter_idxs])
streamgraph.format_streamgraph(ax)
ax.xaxis.set_ticks(times)
ax.autoscale_view()
# fig.show()
fig.savefig(args[1],dpi=120)
# sort streamgraphs appropriately
argsort_onset = streamgraph.argsort_onset(streams)
streams = streams[argsort_onset]
matrix = matrix[argsort_onset]
colors = colors[argsort_onset]
# argsort_inside_out = streamgraph.argsort_inside_out(streams)
# streams = streams[argsort_inside_out]
# colors = colors[argsort_inside_out]
# filter out some clones
if options.filter == 'none':
filter_idxs = np.ones(streams.shape[0]) > 0 # all streams
elif options.filter == 'seen2':
filter_idxs = np.sum(streams > 0, axis=1) >= 2 # seen twice
elif options.filter == 'sum2':
filter_idxs = np.sum(matrix, axis=1) >= 2 # sum=2
elif options.filter == 'sum3':
filter_idxs = np.sum(matrix, axis=1) >= 3 # sum=3
else:
raise ValueError, "what filter do you want me to use?"
fig = plt.figure(figsize=(24,16))
ax = fig.add_subplot(111)
streamgraph.streamgraph(ax, streams[filter_idxs], x=times, colors=colors[filter_idxs])
streamgraph.format_streamgraph(ax)
ax.xaxis.set_ticks(times)
ax.autoscale_view()
# fig.show()
fig.savefig(args[1],dpi=120)
