Intellipaat Back

Explore Courses Blog Tutorials Interview Questions
0 votes
2 views
in Machine Learning by (19k points)

I'm trying to plot multiple series with two measurements (so it's actually num_of_time_series x 2 graphs) in one figure using pygal. For instance, suppose my data is:

from collections import defaultdict

measurement_1=defaultdict(None,[

  ("component1", [11.83, 11.35, 0.55]), 

  ("component2", [2.19, 2.42, 0.96]),

  ("component3", [1.98, 2.17, 0.17])])

measurement_2=defaultdict(None,[

  ("component1", [34940.57, 35260.41, 370.45]),

  ("component2", [1360.67, 1369.58, 2.69]),

  ("component3", [13355.60, 14790.81, 55.63])])

x_labels=['2016-12-01', '2016-12-02', '2016-12-03']

and the graph rendering code is that:

from pygal import graph

import pygal

def draw(measurement_1, measurement_2 ,x_labels):

  graph = pygal.Line()

  graph.x_labels = x_labels

  for key, value in measurement_1.iteritems():

      graph.add(key, value)

  for key, value in measurement_2.iteritems():

      graph.add(key, value, secondary=True)

  return graph.render_data_uri()

The Current result is that.

The problem in the code above is that it's unclear which graph represents measurement 1 and which represents measurement 2. Second, I would like to see each component in a different color(or shape).

This graph aims to compare one component versus the two others, and to see the correlation between measurement 1 and 2.

Thanks for the help guys!

1 Answer

0 votes
by (33.1k points)

Your problem can be solved by using the following code:

from pygal import graph

import pygal

def draw(measurement_1, measurement_2 ,x_labels):

  graph = pygal.Line()

  graph.x_labels = x_labels

  for key, value in measurement_1.iteritems():

     ##

     if "component1":

        graph.add(key, value, stroke_style={'width': 5, 'dasharray': '3, 6', 'linecap': 'round', 'linejoin': 'round'})

     else:

     ##

        graph.add(key, value)

  for key, value in measurement_2.iteritems():

      graph.add(key, value, secondary=True)

  return graph.render_data_uri()

Hope this answer helps.

31k questions

32.8k answers

501 comments

693 users

Browse Categories

...