In [21]:
import pandas as pd
import numpy as np
import re
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns

C:\Anaconda2\lib\site-packages\matplotlib\__init__.py:872: UserWarning: axes.color_cycle is deprecated and replaced with axes.prop_cycle; please use the latter.
  warnings.warn(self.msg_depr % (key, alt_key))

2014 entrant counts

In [7]:
entrants2014 = pd.read_csv('data/2014_participating_countries.txt', header=None, names=['country', 'entries'])
print entrants2014.dtypes
print entrants2014.shape
entrants2014.head()

country    object
entries     int64
dtype: object
(58, 2)
Out[7]:
country entries
0 Argentina 8
1 Australia 80
2 Austria 32
3 Belgium 89
4 Belize 4
In [14]:
entry_count = np.sum(entrants2014.entries)
entry_count
Out[14]:
4753L
In [16]:
entrants2014['perOfTotal'] = entrants2014.entries/entry_count
In [38]:
topTen2014 = entrants2014.sort_values('entries', axis=0, ascending=False).head(10)
topTen2014.reset_index(drop=True)
Out[38]:
country entries perOfTotal
0 United States 3402 0.715758
1 Germany 249 0.052388
2 Canada 190 0.039975
3 Belgium 89 0.018725
4 Australia 80 0.016831
5 Japan 71 0.014938
6 United Kingdom 69 0.014517
7 Brazil 63 0.013255
8 Italy 63 0.013255
9 New Zealand 50 0.010520
In [39]:
topTen2014
Out[39]:
country entries perOfTotal
56 United States 3402 0.715758
21 Germany 249 0.052388
8 Canada 190 0.039975
3 Belgium 89 0.018725
1 Australia 80 0.016831
31 Japan 71 0.014938
55 United Kingdom 69 0.014517
6 Brazil 63 0.013255
30 Italy 63 0.013255
38 New Zealand 50 0.010520
In [20]:
np.sum(topTen2014.perOfTotal)
Out[20]:
0.910162002945508
In [41]:
plt.figure(figsize=(12,8))
sns.plt.title("2014 World Beer Cup Top 10 Country Entrants")
sns.barplot(topTen2014.country, topTen2014.entries, palette="BuGn_d")
sns.plt.xlabel("Country")
sns.plt.ylabel("Entries")
sns.plt.savefig('img/wbc_2014_top10entrants.png')

2014 awards by country

In [46]:
winners2014 = pd.read_csv('data/2014_award_counts_country.txt', header=None, names=['country', 'awards'])
print winners2014.dtypes
winners2014['perOfTot'] = winners2014.awards/np.sum(winners2014.awards)
print winners2014.shape
winners2014.sort_values('awards', axis=0, ascending=False)

country    object
awards      int64
dtype: object
(22, 3)
Out[46]:
country awards perOfTot
21 United States 205 0.729537
7 Germany 27 0.096085
4 Canada 7 0.024911
2 Belgium 5 0.017794
20 United Kingdom 5 0.017794
10 Italy 4 0.014235
14 Netherlands 3 0.010676
0 Australia 3 0.010676
5 Czech Republic 3 0.010676
11 Japan 3 0.010676
6 El Salvador 2 0.007117
1 Austria 2 0.007117
15 New Zealand 2 0.007117
3 Brazil 2 0.007117
8 Ireland 1 0.003559
9 Israel 1 0.003559
12 Latvia 1 0.003559
13 Mexico 1 0.003559
16 Poland 1 0.003559
17 South Korea 1 0.003559
18 Sweden 1 0.003559
19 Taiwan 1 0.003559
In [47]:
winners2014.sort_values('awards', axis=0, ascending=False).head(10)
Out[47]:
country awards perOfTot
21 United States 205 0.729537
7 Germany 27 0.096085
4 Canada 7 0.024911
2 Belgium 5 0.017794
20 United Kingdom 5 0.017794
10 Italy 4 0.014235
14 Netherlands 3 0.010676
0 Australia 3 0.010676
5 Czech Republic 3 0.010676
11 Japan 3 0.010676
In [49]:
np.sum(winners2014.sort_values('awards', axis=0, ascending=False).head(10)['perOfTot'])
Out[49]:
0.9430604982206405
In [51]:
WBC = pd.read_csv('data/wbc_PDF98_14c.csv')
WBC.tail()
Out[51]:
Unnamed: 0 award beer brewery category origin year city country_state country category_num entries category_name
2266 NaN Bronze Veritas 002 Port Brewing Co. and The Lost Abbey Category 13: Experimental Beer (Lager or Ale),... San Marcos, CA 2008 San Marcos CA USA Category 13 27 Experimental Beer (Lager or Ale)
2267 NaN Silver Hopsquatch '06 Four Peaks Brewing Co. Category 18: Aged Beer (Ale or Lager), 25 Entries Tempe, AZ 2008 Tempe AZ USA Category 18 25 Aged Beer (Ale or Lager)
2268 NaN Gold Alaskan Smoked Porter 2007 Alaskan Brewing Co. Category 20: Smoke-Flavored Beer, 27 Entries Juneau, AK 2008 Juneau AK USA Category 20 27 Smoke-Flavored Beer
2269 NaN Gold Saint Arnold Divine Reserve No. 4 Saint Arnold Brewing Co Category 64: Strong Scotch Ale, 43 Entries Houston, TX 2008 Houston TX USA Category 64 43 Strong Scotch Ale
2270 NaN Silver Hopsquatch '07 Four Peaks Brewing Co. Category 68: Barley Wine-Style Ale, 61 Entries Tempe, AZ 2008 Tempe AZ USA Category 68 61 Barley Wine-Style Ale
In [56]:
usa2014 = WBC[(WBC.country=="USA") & (WBC.year==2014)]
print usa2014.shape
usa2014.head()

(202, 13)
Out[56]:
Unnamed: 0 award beer brewery category origin year city country_state country category_num entries category_name
1939 1940 Gold Float Trip Piney River Brewing Co. Category 1: American-Style Wheat Beer, 23 Entries Bucyrus, MO 2014 Bucyrus MO USA Category 1 23 American-Style Wheat Beer
1940 1941 Silver Red Dawn Carson�s Brewery Category 1: American-Style Wheat Beer, 23 Entries Evansville, IN 2014 Evansville IN USA Category 1 23 American-Style Wheat Beer
1941 1942 Bronze DD Blonde Hop Valley Brewing Co. Category 1: American-Style Wheat Beer, 23 Entries Eugene, OR 2014 Eugene OR USA Category 1 23 American-Style Wheat Beer
1942 1943 Gold Hefeweizen Widmer Brothers Brewing Co. Category 2: American-Style Wheat Beer With Yea... Portland, OR 2014 Portland OR USA Category 2 25 American-Style Wheat Beer With Yeast
1943 1944 Silver White Noise �berbrew Category 2: American-Style Wheat Beer With Yea... Billings, MT 2014 Billings MT USA Category 2 25 American-Style Wheat Beer With Yeast
In [60]:
states2014 = usa2014.groupby('country_state')
In [67]:
states2014.award.count().sort_values(ascending=False).head()
Out[67]:
country_state
CA    36
CO    22
OR    17
IL     8
WA     8
Name: award, dtype: int64

IBM data science workbench, "Tutorial #2 - Explore & Visualize"

In [70]:
WBC.award.value_counts()
Out[70]:
Silver    764
Bronze    754
Gold      753
Name: award, dtype: int64
In [72]:
award_by_country = WBC.groupby(['country', 'award']).size()
award_by_country.head(12)
Out[72]:
country    award 
Argentina  Bronze     1
           Gold       2
           Silver     1
Australia  Bronze    13
           Gold      14
           Silver     9
Austria    Bronze     4
           Gold       5
           Silver     7
Belgium    Bronze    21
           Gold      37
           Silver    24
dtype: int64
In [74]:
award_by_country_df = award_by_country.unstack()
award_by_country_df.head()
Out[74]:
award Bronze Gold Silver
country
Argentina 1 2 1
Australia 13 14 9
Austria 4 5 7
Belgium 21 37 24
Bolivia 1 1 NaN
In [75]:
award_by_country_df.fillna(0, inplace=True)
In [76]:
award_by_country_df.idxmax()
Out[76]:
award
Bronze    USA
Gold      USA
Silver    USA
dtype: object
In [77]:
award_by_country_df.sort('Gold', ascending=False, inplace=True)

C:\Anaconda2\lib\site-packages\ipykernel\__main__.py:1: FutureWarning: sort(columns=....) is deprecated, use sort_values(by=.....)
  if __name__ == '__main__':
In [79]:
award_by_country_df = award_by_country_df[['Gold', 'Silver', 'Bronze']]
award_by_country_df.head(10)
Out[79]:
award Gold Silver Bronze
country
USA 534 529 532
Germany 63 67 64
Belgium 37 24 21
Canada 21 35 36
Japan 14 23 26
Australia 14 9 13
UK 11 8 6
Czech Republic 7 8 6
Singapore 5 1 1
Austria 5 7 4
In [89]:
award_by_country_df[:10].plot(kind='bar', figsize=(12,10), colormap='winter')
Out[89]:
<matplotlib.axes._subplots.AxesSubplot at 0xe3579b0>
In [119]:
plot_colors = ['gold', 'silver', 'saddlebrown']
award_by_country_df[:10].plot(kind='bar', figsize=(12,10), color=plot_colors)
Out[119]:
<matplotlib.axes._subplots.AxesSubplot at 0x1346ae48>
In [90]:
usa_award_year = WBC[WBC.country=="USA"].groupby('year').size()
usa_award_year
Out[90]:
year
1996     99
1998    118
2000    118
2002    141
2004    157
2006    159
2008    190
2010    203
2012    208
2014    202
dtype: int64
In [91]:
usa_award_year.plot(kind='bar', figsize=(12,6))
Out[91]:
<matplotlib.axes._subplots.AxesSubplot at 0xe37c0f0>

IBM data science workbench, "Tutorial #4 - Analyze Data"

In [92]:
from IPython.core.pylabtools import figsize
plt.rcParams['axes.facecolor'] = '#F5F5F5'

figsize(11, 6)

import scipy.stats as stats
a = np.arange(30)
poi = stats.poisson
lambdas = [2.5, 14.5]
colors = ["#6B6A96", "#943A45"]

# plot a Poisson distribution for each lambda
for lambda_, color in zip(lambdas, colors):
    plt.bar(a, poi.pmf(a, lambda_), color=color,
            label="$\lambda = %.1f$" % lambda_, alpha=0.50,
            edgecolor=color, lw="1.5")

plt.title("Probability mass function of a Poisson random variable")
plt.legend()
plt.grid()
plt.ylabel("probability of $n$")
plt.xticks(a + 0.4, a)
plt.xlabel("$n$")
plt.show()
In [95]:
# conda install -c https://conda.binstar.org/pymc pymc
In [96]:
import pymc as pm
import numpy as np

def make_model(data):
    '''
    Create a model for use in predicting lambda_1, lambda_2, and tau,
    which combine to create our posterior probability distributions
    (one before tau, and one after tau).
    '''
    # prior lambda estimates
    alpha = 1.0 / data.mean() 
    lambda_1 = pm.Exponential("lambda_1", alpha)
    lambda_2 = pm.Exponential("lambda_2", alpha)
    # prior tau estimates
    tau = pm.DiscreteUniform("tau", lower=0, upper=len(data))

    @pm.deterministic
    def lambda_(tau=tau, lambda_1=lambda_1, lambda_2=lambda_2):
        '''Create random value of lambda, based on prior lambda.'''
        out = np.zeros(len(data))
        out[:tau] = lambda_1  # lambda before tau
        out[tau:] = lambda_2  # lambda after (and including) tau
        return out
    
    # lambda is Poisson distributed
    observation = pm.Poisson("obs", lambda_, value=data.values, observed=True)

    model = pm.Model([observation, lambda_1, lambda_2, tau])
    return model
In [97]:
model = make_model(usa_award_year)
In [98]:
mcmc = pm.MCMC(model)
mcmc.sample(50000, 10000, 1)

 [-----------------100%-----------------] 50000 of 50000 complete in 10.7 sec
In [99]:
lambda_1_samples = mcmc.trace('lambda_1')[:]
lambda_2_samples = mcmc.trace('lambda_2')[:]
tau_samples = mcmc.trace('tau')[:]
In [121]:
# our data
data = usa_award_year
n_data = len(data)
years = data.index.values

figsize(10, 12)
xticks = np.arange(80, 200, 10)
yticks = np.arange(0.0, 1.1, 0.1)

# lambda_1
ax = plt.subplot(311)
ax.set_autoscaley_on(False)
plt.hist(lambda_1_samples, histtype='stepfilled', bins=30, alpha=0.65,
         label="posterior of $\lambda_1$", color="#6B6A96", normed=True)
plt.legend(loc="upper left")
plt.xticks(xticks)
# plt.yticks(yticks)
plt.ylim([0., .06])
plt.xlabel("$\lambda_1$ value")
plt.ylabel("probability");
plt.title(r"""Posterior distributions of $\lambda_1,\;\lambda_2,\;\tau$""")
plt.grid()

# lambda_2
ax = plt.subplot(312)
ax.set_autoscaley_on(False)
plt.hist(lambda_2_samples, histtype='stepfilled', bins=30, alpha=0.65,
         label="posterior of $\lambda_2$", color="#943A45", normed=True)
plt.legend(loc="upper left")
plt.xticks(xticks)
# plt.xlim([xticks.min(), xticks.max()])
plt.xlim([150, 250])
# plt.yticks(yticks)
plt.ylim([0., .06])
plt.xlabel("$\lambda_2$ value")
plt.ylabel("probability");
plt.grid()

# tau
ax = plt.subplot(313)
w = 1.0 / tau_samples.shape[0] * np.ones_like(tau_samples)
plt.hist(tau_samples, bins=n_data, alpha=1,
         label=r"posterior of $\tau$",
         color="#799631", weights=w, rwidth=2.)
plt.legend(loc="upper left")
plt.xticks(np.arange(n_data), years, rotation='vertical')
plt.yticks(yticks)
plt.xlabel(r"$\tau$ (year of Cup)")
plt.ylabel("probability");
plt.grid()
In [102]:
# plot medals awarded per games
ax = WBC.groupby('year').size()\
.plot(figsize=(10,5), legend=True, label='Medals')
ax.set_ylabel('award')

# plot events per games
events_year = WBC.drop_duplicates(cols=['year', 'category_name'])\
.groupby('year').size()
events_year.plot(secondary_y=True, legend=True, label='Events', mark_right=False)
ax.right_ax.set_ylabel('Events')

C:\Anaconda2\lib\site-packages\ipykernel\__main__.py:6: FutureWarning: the 'cols' keyword is deprecated, use 'subset' instead
Out[102]:
<matplotlib.text.Text at 0xf63e9b0>

category entries heatmap

In [ ]: