Author | ben haley |
Submission date | 2012-08-11 04:53:50.017642 |
Rating | 4668 |
Matches played | 750 |
Win rate | 47.6 |
Use rpsrunner.py to play unranked matches on your computer.
"""
Using baysian logic and laplace smoothing to predict
the most likely next move. Plays only conservative
moves.
bmh July 2012 <benjamin.haley@gmail.com>
"""
from random import choice
moves = 'RPS'
values = {
'RR': 0.,
'RP': 1.,
'RS': -1.,
'PR': -1.,
'PP': 0.,
'PS': 1.,
'SR': 1.,
'SP': -1.,
'SS': 0.,
}
def get_probable_moves(history, laplace=1, lookback=6):
odds = {'R':1./3, 'P':1./3, 'S':1./3}
h, l = history, laplace
for i in range(1, lookback + 1):
found = h.count(h[-i:])
for m in moves:
odds[m] *= (1.*(h.count(h[-i:] + m) + l) / (found + l))
normalize = sum(odds.values())
for m in moves:
odds[m] /= normalize
return odds
def get_move_values(probs):
"""return the value of each move, RPS, given a
dictionary of the odds that the oponent will make
any given move"""
values_ = {}
for my in moves:
values_[my] = sum([p * values[his+my] for his, p in probs.items()])
return values_
def player(history):
odds = get_probable_moves(history, laplace, lookback)
values_ = get_move_values(odds)
if max(values_.values()) > threshold:
print round(max(values_.values()), 2)
return max(values_, key=values_.get)
else:
print '.',
return choice(moves)
#"""
# to compare with existing code at
# http://www.rpscontest.com/submit
laplace = 5.
threshold = 0.20
lookback = 1
if input == '':
history = ''
else:
history += input
output = player(history)
history += output
def test():
assert get_move_values({'R':1., 'P':0., 'S':0.})=={'P': 1.0, 'S': -1.0, 'R': 0.0}
assert get_move_values({'R':0.5, 'P':0.5, 'S':0.}) == {'P': 0.5, 'S': 0.0, 'R': -0.5}
assert get_probable_moves('R', 0, 0)['R'] == 1./3
assert get_probable_moves('RR', 0, 1)['R'] == 1.
assert get_probable_moves('RR', 1, 1)['R'] == 2. / 4
assert get_probable_moves('RP'*5)['R'] > get_probable_moves('RP'*5)['S']
assert player('RRRRRR') == 'P'
#test()