back to blog

Below, I generate two plots using data from the Pokémon Database. I used data on Generation I Pokémon species base stats, as individual Pokémon can possess variable base stats. Total, as depicted in the first chart, is the sum of HP, Attack, Defense, Special Attack, Special Defense, and Speed.

from bs4 import BeautifulSoup
import urllib.request
import pandas as pd
import numpy as np
from __future__ import division
from bokeh.models import HoverTool
from bokeh.plotting import ColumnDataSource, figure, show
from bokeh.io import output_notebook
from bokeh.models import Circle
from bokeh.resources import CDN
output_notebook()
%matplotlib inline

Loading BokehJS ...

url1 = "http://pokemondb.net/pokedex/stats/height-weight" # pokemon ht/wt database
req = urllib.request.Request(url1, headers={'User-Agent' : "Magic Browser"}) 
page1 = urllib.request.urlopen(req)
soup1 = BeautifulSoup(page1, "html.parser")
table1 = soup1.find("table")

pokemon = []
ht = []
wt = []
bmi = []

# scrape the gen 1 pokemon
counter = 1
for row in table1.findAll("tr"):
    cells = row.findAll("td")
    if len(cells) >=4 and len(cells[1].findAll(text=True)) == 1 and counter < 152:
        pokemon.append(cells[1].findAll(text=True)[0])
        ht.append(cells[3].findAll(text=True)[0]) 
        wt.append(cells[4].findAll(text=True)[0]) 
        bmi.append(cells[5].findAll(text=True)[0])
        counter += 1

pokedex1 = pd.DataFrame({'pokemon': pokemon,
                        'ht': ht,
                        'wt': wt,
                        'bmi': bmi})

def get_ht_wt(x):
    '''
    function for formatting data
    '''
    return float(x.split('(')[1].split(')')[0].split('m')[0].split()[0])

# format the data
for col in pokedex1:
    if col == 'ht' or col == 'wt':
        pokedex1[col] = pokedex1[col].apply(get_ht_wt)
    elif col == 'bmi':
        pokedex1[col] = pokedex1[col].apply(float)

url2 = "http://pokemondb.net/pokedex/all" # pokemon main stats
req = urllib.request.Request(url2, headers={'User-Agent' : "Magic Browser"}) 
page2 = urllib.request.urlopen(req)
soup2 = BeautifulSoup(page2, "html.parser")
table2 = soup2.find("table")

pokemon = []
total = []

# scrape the gen 1 pokemon
counter = 1
for row in table2.findAll("tr"):
    cells = row.findAll("td")
    if len(cells) >=4 and len(cells[1].findAll(text=True)) == 1 and counter < 152:
        pokemon.append(cells[1].findAll(text=True)[0])
        total.append(cells[3].findAll(text=True)[0]) 
        counter += 1

pokedex2 = pd.DataFrame({'pokemon': pokemon,
                        'total': total
                       })

# group pokemon by evolution group
evolution = [1,1,1,2,2,2,3,3,3,4,4,4,5,5,5,6,6,6,7,7,8,8,9,9,10,10,11,11,12,12,12,13,13,
13,14,14,15,15,16,16,17,17,18,18,18,19,19,20,20,21,21,22,22,23,23,24,24,25,25,26,26,26,27,
27,27,28,28,28,29,29,29,30,30,31,31,31,32,32,33,33,34,34,35,36,36,37,37,38,38,39,39,40,40,
40,41,42,42,43,43,44,44,45,45,46,46,'47a','47b',48,49,49,50,50,51,52,53,54,54,55,55,56,56,57,58,
59,60,61,62,63,64,64,65,66,67,67,67,67,68,69,69,70,71,72,73,74,75,76,77,77,77,78,79]
pokedex1['evolution'] = pd.Series(evolution)

# merge pokedexes
merged = pd.merge(pokedex1,pokedex2)
merged['total'] = merged['total'].apply(float)

# generate separate dfs for each evolution group
groups = {}
for i in merged['evolution'].unique():
    groups[i] = merged[merged['evolution'] == i].copy()
    
groups2 = {}
for i in groups:
    groups[i].index = groups[i]['pokemon']
    j = groups[i].index[0]
    groups2[j] = groups[i].copy()
    groups2[j] = groups2[j]['total']
    groups2[j] = pd.DataFrame(groups2[j]).T
    if i == 67:
        groups2[j] = groups2[j][groups2[j].columns[:2]]
    groups2[j].index = [j]
    if len(groups2[j].T) < 3:
        while len(groups2[j].T) < 3:
            groups2[j][len(groups2[j].T)+1] = np.nan
    groups2[j].columns = [1,2,3]

# a df with first evolution as index and 3 columns for each evolution
df = groups2[sorted(groups2.keys())[0]]
for i in sorted(groups2.keys()):
    if i != sorted(groups2.keys())[0]:
        df = df.append(groups2[i])

# create list of pokemon names for plotting
names = []
for i in sorted(groups2.keys()):
    for j in groups.keys():
        for k in groups[j]['pokemon']:
            if i == k:
                if i == 'Eevee':
                    names.append('Eevee')
                    names.append('Flareon/Jolteon/Vaporeon')
                    names.append('N/A')
                else:
                    counter = 0
                    for l in groups[j]['pokemon']:
                        names.append(groups[j]['pokemon'][l])
                        counter += 1
                    while counter < 3:
                        names.append('N/A')
                        counter += 1

df.replace(np.nan, 0, inplace=True) # replace NaN's w/ 0 for plotting purposes

percentiles = merged.describe(percentiles = np.arange(1/8, 1, 1/8))['total'] # power percentiles

stages = ['1','2','3'] # evolution stages
pokemons = list(df.index) # evolution 1 pokemons
colors = ["#d9d9d9", "#deebf7", "#c6dbef", "#9ecae1", "#6baed6", "#4292c6", "#2171b5", "#08519c", "#08306b"] # colormap

# obtain data for plot
stage = []
pokemon = []
color = []
power = []
for p in pokemons:
    for s in stages:
        stage.append(s)
        pokemon.append(p)
        value = int(df[int(s)][p])
        if value == 0:
            power.append('N/A')
        else:
            power.append(value)
        if value == 0:
            color.append(colors[0])
        elif value > 0 and value <= percentiles[4]:
            color.append(colors[1])
        elif value > percentiles[4] and value <= percentiles[5]:
            color.append(colors[2])
        elif value > percentiles[5] and value <= percentiles[6]:
            color.append(colors[3])
        elif value > percentiles[6] and value <= percentiles[7]:
            color.append(colors[4])
        elif value > percentiles[7] and value <= percentiles[8]:
            color.append(colors[5])
        elif value > percentiles[8] and value <= percentiles[9]:
            color.append(colors[6])
        elif value > percentiles[9] and value <= percentiles[10]:
            color.append(colors[7])
        elif value > percentiles[10] and value <= percentiles[11]:
            color.append(colors[8])
            
source = ColumnDataSource(
    data=dict(stage=stage, pokemon=pokemon, color=color, power=power, names=names)
)

TOOLS = "hover,save,pan,box_zoom,wheel_zoom"

p = figure(title="Pokémon Total Chart",
           x_range=stages, y_range=list(reversed(pokemons)),
           x_axis_location="above", plot_width=400, plot_height=900,
           tools=TOOLS)

p.grid.grid_line_color = None
p.axis.axis_line_color = None
p.axis.major_tick_line_color = None
p.axis.major_label_text_font_size = "6pt"
p.axis.major_label_standoff = 0
p.xaxis.axis_label = "Evolution"

p.rect("stage", "pokemon", 1, 1, source=source,
       color="color", line_color=None)

p.select_one(HoverTool).tooltips = [
    ('Pokemon', '@names'),
    ('Power', '@power'),
]

show(p)

source = ColumnDataSource(
    data=dict(ht=list(merged["ht"]), wt=list(merged["wt"]), pokemon=list(merged["pokemon"]))
)

p = figure(title="Pokémon Height vs. Weight", tools="tap,hover,box_zoom,wheel_zoom")
p.circle("wt", "ht", source = source, size=10, name="circle", alpha=.2, color="black", line_color=None)

p.select_one(HoverTool).tooltips = [
    ('Pokemon', '@pokemon'),
]

selected_circle = Circle(fill_alpha=1, fill_color="firebrick", line_color=None)
nonselected_circle = Circle(fill_alpha=0.2, fill_color="black", line_color=None)

renderer = p.select(name="circle")
renderer.selection_glyph = selected_circle
renderer.nonselection_glyph = nonselected_circle

p.xaxis.axis_label = "Wt (kg)"
p.yaxis.axis_label = "Ht (m)"
show(p)