pokemon-datascience

import pandas as pd pokemon_data = pd.read_csv('pokemon.csv') pokemon_data = pokemon_data.drop(['japanese_name'], axis = 1) # We don't need japanese_name column for our analysis pokemon_data.head(5)

	abilitiesobject	against_bugfloat64	against_darkfloat64	against_dragonfloat64	against_electricfloat64	against_fairyfloat64	against_fightfloat64	against_firefloat64	against_flyingfloat64	against_ghostfloat64	against_grassfloat64	against_groundfloat64	against_icefloat64	against_normalfloat64	against_poisonfloat64	against_psychicfloat64	against_rockfloat64	against_steelfloat64	against_waterfloat64	attackint64	base_egg_stepsint64	base_happinessint64	base_totalint64	capture_rateobject	classficationobject	defenseint64	experience_growthint64	height_mfloat64	hpint64	nameobject	percentage_malefloat64	pokedex_numberint64	sp_attackint64	sp_defenseint64	speedint64	type1object	type2object	weight_kgfloat64	generationint64	is_legendaryint64
0	['Overgrow', 'Chlorophyll']	1	1	1	0.5	0.5	0.5	2	2	1	0.25	1	2	1	1	2	1	1	0.5	49	5120	70	318	45	Seed Pokémon	49	1059860	0.7	45	Bulbasaur	88.1	1	65	65	45	grass	poison	6.9	1	0
1	['Overgrow', 'Chlorophyll']	1	1	1	0.5	0.5	0.5	2	2	1	0.25	1	2	1	1	2	1	1	0.5	62	5120	70	405	45	Seed Pokémon	63	1059860	1	60	Ivysaur	88.1	2	80	80	60	grass	poison	13	1	0
2	['Overgrow', 'Chlorophyll']	1	1	1	0.5	0.5	0.5	2	2	1	0.25	1	2	1	1	2	1	1	0.5	100	5120	70	625	45	Seed Pokémon	123	1059860	2	80	Venusaur	88.1	3	122	120	80	grass	poison	100	1	0
3	['Blaze', 'Solar Power']	0.5	1	1	1	0.5	1	0.5	1	1	0.5	2	0.5	1	1	1	2	0.5	2	52	5120	70	309	45	Lizard Pokémon	43	1059860	0.6	39	Charmander	88.1	4	60	50	65	fire	nan	8.5	1	0
4	['Blaze', 'Solar Power']	0.5	1	1	1	0.5	1	0.5	1	1	0.5	2	0.5	1	1	1	2	0.5	2	64	5120	70	405	45	Flame Pokémon	58	1059860	1.1	58	Charmeleon	88.1	5	80	65	80	fire	nan	19	1	0
5 rows × 40 columns

# lets grab some visualization libraries for some simple visualizations of our data import matplotlib.pyplot as plt import seaborn as sns # grab only the name column for each generation so we have a definitive count poke_by_generation = pokemon_data.groupby('generation').count().name plt.figure(figsize=(10, 10)) bars = plt.bar(range(1 , len(poke_by_generation) + 1), poke_by_generation, width=0.7) # add some formatting to our plot plt.ylabel("number of pokemon") plt.xlabel("generation") plt.title(label="Pokemon Per Generation", loc="center", fontsize=22) for bar in bars: yval = bar.get_height() plt.text(x = bar.get_x()+0.25, y = yval, s= yval, va='bottom', fontsize=12) # bottom vertical alignment so the number doesn't touch the bar plt.show()

columns_of_interest = ['height_m', 'weight_kg', 'hp', 'attack', 'sp_attack', 'defense', 'sp_defense', 'speed'] corr_data = pokemon_data[columns_of_interest] corr_data = corr_data.dropna() import numpy as np correlation_coefficients = np.corrcoef( corr_data, rowvar=False ) plt.figure(figsize=(15,8)) sns.heatmap( correlation_coefficients[0:2], annot=True ) # index from 0:2 because we only want to see corrolations between height and weight plt.xticks( np.arange(8)+0.5, corr_data.columns, rotation = 35 ) plt.yticks( np.arange(2)+0.5, corr_data.columns[0:2], rotation = 0 ) plt.title(label="Pokemon Stats Correlated With Height & Weight", loc="center", fontsize=22) plt.show()

pokemon_data.loc[pokemon_data['weight_kg'].isna()][['name', 'generation']]

	nameobject Rattata5% Raticate5% 18 others90%	generationint641 – 7
18	Rattata	1
19	Raticate	1
25	Raichu	1
26	Sandshrew	1
27	Sandslash	1
Expand rows 5 - 14
88	Muk	1
102	Exeggutor	1
104	Marowak	1
719	Hoopa	6
744	Lycanroc	7
20 rows × 2 columns

legendary = pokemon_data[pokemon_data['is_legendary'] == True] non_legendary = pokemon_data[pokemon_data['is_legendary'] == False]

corr_data_legendary = legendary[columns_of_interest].dropna() corr_data_non_legendary = non_legendary[columns_of_interest].dropna() plt.figure(figsize=(12,8)) correlation_coefficients = np.corrcoef( corr_data_non_legendary, rowvar=False ) sns.heatmap( correlation_coefficients[0:2], annot=True ) # index from 0:2 because we only want to see corrolations between height and weight plt.xticks( np.arange(8)+0.5, corr_data.columns, rotation = 35 ) plt.yticks( np.arange(2)+0.5, corr_data.columns[0:2], rotation = 0 ) plt.title(label="Non-Legendary Pokemon Stats Correlated With Height & Weight", loc="center", fontsize=22) plt.show() plt.figure(figsize=(12,8)) correlation_coefficients = np.corrcoef( corr_data_legendary, rowvar=False ) sns.heatmap( correlation_coefficients[0:2], annot=True ) # index from 0:2 because we only want to see corrolations between height and weight plt.xticks( np.arange(8)+0.5, corr_data.columns, rotation = 35 ) plt.yticks( np.arange(2)+0.5, corr_data.columns[0:2], rotation = 0 ) plt.title(label="Legendary Pokemon Stats Correlated With Height & Weight", loc="center", fontsize=22) plt.show()

import requests # we will use the requests library to call the pokeapi import json # the response will be easier to work with in JSON, so we opt for the library # loop through all of our pokedex entries for pokedex_entry in range(1, pokemon_data.shape[0]): # make a request to the amazing pokeapi! request_url = 'https://pokeapi.co/api/v2/pokemon/' + str(pokedex_entry) response = requests.get(request_url) # make our JSON response into a python dictionary for ease of mutability full_response = json.loads(response.text) image_url = full_response['sprites']['front_default'] # extract the default image URL from the dictionary pokemon_data.loc[pokedex_entry -1, 'image_url'] = image_url # append to the correct index in the dataframe

# Exporting the csv to be used in our dashboard application pokemon_data.to_csv('pokemon_data.csv')