import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
%pylab inline

source = 'https://gist.githubusercontent.com/chengsoonong/dede21b2eefa43b30d14/raw/add9312ea30c0ccc770151cb6188f964d2baa047/gsoc2014_accepted.csv'
data = pd.read_table(source, sep=',')
print(data.columns.values)
orgs = np.unique(data['Organization'])
proj_per_org = []
for org in orgs:
    proj_per_org.append(len(np.flatnonzero(data['Organization'] == org)))
proj_per_org = np.array(proj_per_org)
sort_idx = np.argsort(-proj_per_org)

fig = plt.figure(figsize=(15,5))
ax = fig.add_subplot(111)
top_orgs = 20
idx = np.arange(top_orgs)
ax.bar(idx, proj_per_org[sort_idx][:top_orgs])
ax.set_xticks(idx)
dummy = ax.set_xticklabels(orgs[sort_idx], rotation=80, ha='center')

ml_orgs = []
for org in orgs:
    if 'learning' in org.lower():
        ml_orgs.append(org)

ml_titles = []
for org in ml_orgs:
    title_from_org = data['Title'][data['Organization']==org]
    for title in title_from_org:
        ml_titles.append(np.flatnonzero(data['Title']==title)[0])

titles = data['Title']
for idx,title in enumerate(titles):
    if ('learning' in title.lower()) or ('scikit-learn' in title.lower()):
        ml_titles.append(np.flatnonzero(data['Title']==title)[0])
        ml_orgs.append(data['Organization'][idx])


ml_orgs = np.unique(np.array(ml_orgs))
print(ml_orgs)

ml_titles = np.unique(np.array(ml_titles))
print(data.loc[ml_titles][['Title','Organization','Student','Mentors']])

ml_proj_count = []
for org in ml_orgs:
    ml_proj_count.append(len(np.flatnonzero(data.loc[ml_titles]['Organization'] == org)))
ml_proj_count = np.array(ml_proj_count)
ml_orgs = np.array(ml_orgs)
sort_idx = np.argsort(-ml_proj_count)

fig = plt.figure(figsize=(10,5))
ax = fig.add_subplot(111)
idx = np.arange(len(ml_orgs))
ax.bar(idx, ml_proj_count[sort_idx])
ax.set_xticks(idx)
dummy = ax.set_xticklabels(ml_orgs[sort_idx], rotation=80)