import numpy as np
import pandas as pd

figsize(15, 5)

df = pd.read_csv('crunchbase.csv')

df.head(5)

df = df.dropna(subset=['founded_year', 'category_code'])

df = df[ (df['founded_year'] >= 2000) & (df['founded_year'] <= 2013) ]

df = df[ df['region'] != 'unknown' ]

df['region'].value_counts().head(10).plot(kind='bar')

df.groupby('region')['name'].count().order(ascending=False).head(5)

num_companies = df.groupby('founded_year')['name'].count()
total_funding = df.groupby('founded_year')['funding_total_usd'].sum()

pd.DataFrame({ 'num_companies' : num_companies, 'total_funding' : total_funding }).plot(secondary_y='total_funding')

sf_funding = df[df['region'] == 'SF Bay'].groupby('founded_year')['funding_total_usd'].sum()
bos_funding = df[df['region'] == 'Boston'].groupby('founded_year')['funding_total_usd'].sum()
ny_funding = df[df['region'] == 'New York'].groupby('founded_year')['funding_total_usd'].sum()

(sf_funding / total_funding).plot()
(bos_funding / total_funding).plot()
(ny_funding / total_funding).plot()