Very newbie question:
I need to draw a bar plot from a list of tuples. The first element is a name (categorical) for the x axis, the second element is float type (for the y axis). I'd also like to order the bars in descending order, and add a trendline. Here is some sample code:
In [20]: popularity_data
Out[20]:
[('Unknown', 10.0),
(u'Drew E.', 240.0),
(u'Anthony P.', 240.0),
(u'Thomas H.', 220.0),
(u'Ranae J.', 150.0),
(u'Robert T.', 120.0),
(u'Li Yan M.', 80.0),
(u'Raph D.', 210.0)]
If you have a list of tuples, you can try the below code to get what you want.
import numpy as np
import matplotlib.pyplot as plt
popularity_data = [('Unknown', 10.0),
(u'Drew E.', 240.0),
(u'Anthony P.', 240.0),
(u'Thomas H.', 220.0),
(u'Ranae J.', 150.0),
(u'Robert T.', 120.0),
(u'Li Yan M.', 80.0),
(u'Raph D.', 210.0)]
# sort in-place from highest to lowest
popularity_data.sort(key=lambda x: x[1], reverse=True)
# save the names and their respective scores separately
# reverse the tuples to go from most frequent to least frequent
people = zip(*popularity_data)[0]
score = zip(*popularity_data)[1]
x_pos = np.arange(len(people))
# calculate slope and intercept for the linear trend line
slope, intercept = np.polyfit(x_pos, score, 1)
trendline = intercept + (slope * x_pos)
plt.plot(x_pos, trendline, color='red', linestyle='--')
plt.bar(x_pos, score,align='center')
plt.xticks(x_pos, people)
plt.ylabel('Popularity Score')
plt.show()
This will give you a plot like the one below, although it doesn't make sense to plot a trend line on a bar plot when you aren't using a time series.
References:
people = zip(*popularity_data)[0] by people = list(zip(*popularity_data))[0]. Indeed, zip returns an iterable in python 3 and not a list.You should use a dictionary, it's easier to use. This gets you the bars in descending order:
popularity_data = {
'Unknown': 10.0,
u'Drew E.': 240.0,
u'Anthony P.': 240.0,
u'Thomas H.': 220.0,
u'Ranae J.': 150.0,
u'Robert T.': 120.0,
u'Li Yan M.': 80.0,
u'Raph D.': 210.0
}
for y in reversed(sorted(popularity_data.values())):
k = popularity_data.keys()[popularity_data.values().index(y)]
print k + ':', y
del popularity_data[k]
You can add the trendline using matplotlib, as Aleksander S suggested.
Also, if you like you can have it stored in a list of tuples as you originally had it like this:
popularity_data = {
'Unknown': 10.0,
u'Drew E.': 240.0,
u'Anthony P.': 240.0,
u'Thomas H.': 220.0,
u'Ranae J.': 150.0,
u'Robert T.': 120.0,
u'Li Yan M.': 80.0,
u'Raph D.': 210.0
}
descending = []
for y in reversed(sorted(popularity_data.values())):
k = popularity_data.keys()[popularity_data.values().index(y)]
descending.append(tuple([k, y]))
del popularity_data[k]
print descending
