def check_bias(times=1000):
"""
Experiment to check if there is a bias in the setting. All buyers have an initial bidding factor random
between 1 and 1.001 and the same increasing and decreasing factor.
The result is presented as the number of times that each buyer wins the simulation.
:param times: Number of times to execute the test
"""
max_profit = np.zeros(n_buyers)
for n in range(times):
auctioneer = Auctioneer(
bidding_factor_strategy=2,
R_rounds=100,
)
auctioneer.bidding_factor = []
for buyer in range(n_buyers):
bid_fact = np.random.uniform(1, 1.001, 3)
auctioneer.bidding_factor.append(bid_fact)
auctioneer.increase_bidding_factor = [1.2 for n in range(n_buyers)]
auctioneer.decrease_bidding_factor = [0.8 for n in range(n_buyers)]
auctioneer.start_auction()
buyers_prof = auctioneer.cumulative_buyers_profits[:, auctioneer.
r_rounds - 1]
for buyer in range(n_buyers):
if buyers_prof[buyer] == max(buyers_prof):
max_profit[buyer] += 1
[
print("Buyer", buyer, "was the one with more profit",
max_profit[buyer], "times") for buyer in range(n_buyers)
]
def check_price_stability_varying_ceiling():
"""
Check the effect of modifying the ceiling for the strategy 2.
The effect is presented with a plot where the x axis is the value of the ciling and the y axis the average
difference between the initial prices and the market price of the last round.
"""
iterations = 500
ceilings = [top for top in np.arange(1.5, 5, 0.1)]
# avg_market_prices = []
# diff_marketprice_start_prices = []
mrkt_price_starting_price_ratio = []
for iter in range(len(ceilings)):
auctioneer = Auctioneer(K_sellers=3,
N_buyers=10,
R_rounds=100,
debug=False,
bidding_factor_strategy=[3 for x in range(10)])
auctioneer.ceiling = ceilings[iter]
auctioneer.start_auction()
mrkt_price_starting_price_ratio.append(
np.mean(auctioneer.market_price[auctioneer.r_rounds - 1]) /
np.mean(auctioneer.starting_prices))
# plt.plot(ceilings, diff_marketprice_start_prices)
plt.plot(ceilings, mrkt_price_starting_price_ratio)
plt.xlabel("Ceiling")
plt.ylabel("Ratio between market price and starting price of last round")
plt.legend()
plt.show()
def test_real_case_scenario(self):
starting_prices = [[40, 50, 20]]
auctioneer = Auctioneer(starting_prices=starting_prices,
M_types=2,
K_sellers=3,
N_buyers=5,
R_rounds=2,
level_comm_flag=False)
auctioneer.increase_bidding_factor = [2, 3, 4, 5, 6]
auctioneer.decrease_bidding_factor = [0.6, 0.5, 0.4, 0.3, 0.2]
auctioneer.sellers_types = [1, 1, 0]
# auctioneer.bidding_factor = np.array([
# # Buyer 0
# [
#
# [1, 2, 3], # Type 0
# [4, 5, 6] # Type 1
# ],
# # Buyer 1
# [
# [1.68791717, 1.43217411, 1.1566692],
# [1.20532547, 1.05372195, 1.19885528]
# ],
# # Buyer 2
# [
# [1.71709178, 1.83604667, 1.4957177],
# [1.50015315, 1.77615324, 1.00780864]
# ],
# # Buyer 3
# [
# [1.62403167, 1.51698165, 1.74709901],
# [1.84536679, 1.29700791, 1.08997174]
# ],
# # Buyer 4
# [
# [1.81391097, 1.2531242, 1.01217679],
# [1.15969576, 1.55215565, 1.34450197]
# ]
# ])
auctioneer.start_auction()
self.assertEqual([], auctioneer.market_price)
"the decreasing factor while if it is lower multiply by the increasing factor.\n"
strategy = request_integer_input(strat)
level_commitment_activated = request_boolean_input(
"Should use level commitment? y/n ")
if level_commitment_activated:
penalty_factor = request_float_input("Penalty factor: ")
else:
penalty_factor = 0
alpha = 0
while alpha != 1 and alpha != 2:
alpha = request_integer_input(
"Bidding factor depends on the sellers (1) or types of items (2): ")
debug = request_boolean_input(
"Print the debug information on every round? (y/n): ")
# Execute with parameters
auctioneer = Auctioneer(
penalty_factor=penalty_factor,
bidding_factor_strategy=[strategy for n in range(number_of_buyers)],
use_seller=(alpha == 1),
M_types=number_of_product_types,
K_sellers=number_of_sellers,
N_buyers=number_of_buyers,
R_rounds=number_of_rounds,
level_comm_flag=level_commitment_activated,
debug=debug)
auctioneer.start_auction()
auctioneer.plot_statistics()
