Cumalitive Gain for addons

gistfile1.txt

def vcg(r, t):
    # r = list of recomendeed addons
    # t = list of gold standard addons from user
    # vanilla cumulative gain    
    r = list(set(filter(bool, r))) # sometimes this needs cleaning due to text formatting issues
    t = list(set(filter(bool, t)))
    
    r_dict = dict(zip(r, [0.0] * len(r)))
    t_dict = dict(zip(t, [1.0 / len(t)] * len(t))) # weights independent of number of groung truth addons
    
    full_dict = dict(r_dict.items() + t_dict.items())
    
    # order is important, ground truth keys in second position overwrite values (0.0) from reccomeded keys in the dict
    return sum([full_dict[x] for x in r]) # score independent of order

# test cases: 
gold_standard_addons = [u'Session Manager', u'ZenMate Security & Privacy VPN', u'Ghostery', u'Hoxx VPN Proxy', u'DoNotTrackMe', u'Adblock Plus', u'Google Reverse Image Search']
recomended_addons = [u'Outlook Notifier', u'Gmail\u2122 Notifier +', u'Black Youtube Theme', u'SQLite Manager', u'DownThemAll!', u'IE View', u'Open in IE\u2122 (Internet Explorer)', u'Quick Preference Button', u'Ghostery']

print vcg(gold_standard_addons,recomended_addons)
print vcg(gold_standard_addons,recomended_addons)
print vcg(recomended_addons, gold_standard_addons)  
print vcg(gold_standard_addons, gold_standard_addons)  
print vcg(recomended_addons, recomended_addons)  
print vcg(recomended_addons[1::4], recomended_addons) 
print vcg(recomended_addons, recomended_addons[1::4])  

print vcg([1,2,3], [1,2,3,4,5,6,7,8,9])
print vcg([1],[1,2])
print vcg([1,2],[1])
print vcg([1,2,3,4,5,6,7],[1])