Home / Questions / 243415
Accepted
Juan
Asked: 2020-03-07 09:40:50 +0800 CST 2020-03-07 09:40:50 +0800 CST

Show information in scatter graph

  • 772

I'm having trouble displaying legends on a scatter.

Data used for the example

import pandas as pd
import numpy as np
import scipy.stats
import matplotlib
import matplotlib.pyplot as pp
import re
import mailbox
import csv
from IPython import display
from ipywidgets import interact, widgets
%matplotlib inline

datos = pd.read_csv('https://raw.githubusercontent.com/theengineeringworld/statistics-using-python/master/gapminder.xls')
datosDos = datos.dropna()
datosDos.reset_index(drop=True, inplace=True)

I am using the following function:

def plotyearDinamico(anio):

    data = datosDos[datosDos.year == anio]
    area = 10e-6 * datosDos.population
    color = data.region.map({'Africa': 'skyblue', 'Europe' : 'gold', 
    'America' : 'palegreen', 'Asia' : 'coral'})

    data.plot.scatter('babies_per_woman', 'age5_surviving', s = area, c = 
    color, linewidth = 1, edgecolors = 'k', figsize=(9,6))

    pp.axis(ymin = 50, ymax = 105, xmin = 0, xmax = 8)
    pp.xlabel('babies_per_woman')
    pp.ylabel('age5_surviving')

Then calling that function I see the following graph:

interact(plotyearDinamico, anio=widgets.IntSlider(min=1950, max=2015, 
step=1, value= 1950))

enter image description here

I know it's a legend() but I can't find how to put it on this chart.

Source: Example of what I need

How can I do to show the names of the continents since I only have them differentiated by colors? Something more or less like this:

import matplotlib.pyplot as plt
from numpy.random import rand

fig, ax = plt.subplots()
for color in ['red', 'green', 'blue']:
    n = 750
    x, y = rand(2, n)
    scale = 200.0 * rand(n)
    ax.scatter(x, y, c=color, s=scale, label=color,
           alpha=0.3, edgecolors='none')

ax.legend()
ax.grid(True)

plt.show()

enter image description here

If I add pp.legend(data.region) to my graph, it shows the following (The area and color match but it only brings one)

enter image description here

pandas

2 Answers

  1. Best Answer

    Various changes to your code:

    def plotyearDinamico(anio):

    data = datosDos[datosDos.year == anio]

    colores = {'Africa': 'skyblue', 'Europe' : 'gold', 
               'America' : 'palegreen', 'Asia' : 'coral'}


    pp.rcParams['figure.figsize'] = [12, 8] # Config. tamaño de gráfica
    fig, ax = pp.subplots()
    for region, color in colores.items():

        # Subgráficas por region
        data_region = data[data.region == region]
        area = 10e-6 * data_region.population
        paises_colors = data_region.region.map(colores)

        ax.scatter(x=data_region.babies_per_woman, 
                   y=data_region.age5_surviving, 
                   c=color, 
                   s=area, 
                   label=region, 
                   linewidth = 1,
                   alpha=0.7,
                   edgecolors='k')

    pp.axis(ymin = 50, ymax = 105, xmin = 0, xmax = 8)
    pp.xlabel('babies_per_woman')
    pp.ylabel('age5_surviving')

    # Configuración del tamaño de las leyendas
    lgnd = pp.legend(loc="lower left", scatterpoints=1, fontsize=10)
    for handle in lgnd.legendHandles:
        handle.set_sizes([50.0])

    pp.grid(True)
    • The main significant change is that, instead of making a single graph, we make sub-graphs by region, for which we will have to make a subset of data, with:data_region = data[data.region == region]
    • We directly use the method scatterof matplotlibso we must properly configure the parameters xe y, by means of:x=data_region.babies_per_woman, y=data_region.age5_surviving
    • In this type of graph it is recommended to configure some transparency through alpha=0.7, to be able to appreciate the overlaps (somewhat at least)
    • When plotting by region, the legend is configured directly using label=region, regionin this case it is a string with its name.
    • by default, the size of the circles adapts to the size of the area of ​​each region, which does not look good, the last part of the code is to set a fixed size.

    enter image description here

    • 1

  2. It seems to me that your code is correct, but the labels are empty, the texts are not defined.

    I'll give you this example: https://www.dkrz.de/up/services/analysis/visualization/sw/ncl/examples/source_code/dkrz-ncl-scatter-plot-with-legend-example

    Where you see that you have to define the texts before displaying it:

    ;-- set explicit labels for legend
    labels = (/"data 1","data 2"/)

    Notice that these labels do appear on the graph:

     pp.xlabel('babies_per_woman')
 pp.ylabel('age5_surviving')

    Now I see more clearly, it seems to me that

    pp.legend(data.region)

    Load a single label and not the array. It should load the entire array.

    • 0