import pandas as pd
import matplotlib.cm
import matplotlib.pyplot as plt
import numpy as np
from string import upper

def bin_pay(city, department, compensation):
    df = pd.DataFrame.from_csv('%s-2012.csv' % city)
    df['job_title'] = df['job_title'].apply(upper)
    dept = df[df['job_title'].str.contains(department.upper())]
    dept['binned_pay'] = dept[compensation].map(lambda x: (int(x) / 10000) * 10000)
    dept = dept[dept['binned_pay'] > 0]
    gb=dept.groupby(by=['binned_pay',])
    mean_pay = dept[compensation].mean()

    print "Average for %(city)s %(department)s $%(mean_pay)2.02f" % locals()
    return gb['binned_pay'].count().apply(lambda n: (100 * ((n*1.0) / len(dept))))

def plot_compare(cities, department, compensation='total_pay_benefits'):
    d = dict((city, bin_pay(city.lower().replace(' ', '-'), department, compensation)) for city in cities)
    df = pd.DataFrame(data=d)
    
    df.plot(kind='bar', colormap='winter', figsize=(18,12), subplots=True)
    return df

df = plot_compare(('Berkeley', 'Albany', 'El Cerrito', 'Richmond', 'Piedmont', 'Oakland', 'Emeryville'), 'police')

df = plot_compare(('Berkeley', 'Albany', 'El Cerrito', 'Richmond', 'Piedmont', 'Oakland', 'Emeryville'), 
                  'police',
                  compensation='base_pay')

df = plot_compare(('Berkeley', 'Albany', 'El Cerrito', 'Richmond', 'Piedmont', 'Oakland', 'Emeryville'), 
                  'police',
                  compensation='total_benefits')