def TrapezTrunc(TC1, TC2, spread1, spread2, N, linf, lsup):
from scipy.stats import trapz
# define variables for trapezoidal distribution
A = TC1 * (1 - spread1)
B = TC2 * (1 + spread2)
c = (TC1 - A) / (B - A)
d = (TC2 - A) / (B - A)
loc = A
scale = B - A
dist = trapz.rvs(c, d, loc, scale, N)
truncdist = [i for i in dist if i >= linf]
truncdist = [i for i in truncdist if i <= lsup]
while len(truncdist) < N:
adddist = trapz.rvs(c, d, loc, scale, N - len(truncdist))
truncadddist = [i for i in adddist if i >= linf]
truncadddist = [i for i in truncadddist if i <= lsup]
truncdist = truncdist + truncadddist
return np.asarray(truncdist)
def TrapezTrunc(TC1, TC2, spread1, spread2, N, linf=float('-inf'), lsup=float('inf')):
if TC1+TC2 == 0:
return np.asarray([0]*N)
# define variables for trapezoidal distribution
if TC1 < TC2:
A = TC1*(1-spread1)
B = TC2*(1+spread2)
c = (TC1-A)/(B-A)
d = (TC2-A)/(B-A)
else:
A = TC2*(1-spread2)
B = TC1*(1+spread1)
c = (TC2-A)/(B-A)
d = (TC1-A)/(B-A)
loc = A
scale = B-A
dist = trapz.rvs(c, d, loc, scale, N)
truncdist = [i for i in dist if i >= linf]
truncdist = [i for i in truncdist if i <= lsup]
while len(truncdist) < N:
adddist = trapz.rvs(c, d, loc, scale, N-len(truncdist))
truncadddist = [i for i in adddist if i >= linf]
truncadddist = [i for i in truncadddist if i <= lsup]
truncdist = truncdist + truncadddist
return np.asarray(truncdist)
def TrapezTrunc(TC1, TC2, spread1, spread2, N, linf, lsup):
if lsup < linf:
raise Exception('lsup should be larger than linf')
if spread1 < 0 or spread2 < 0:
raise Exception('spread should not be below 0')
if TC1 > TC2:
raise Exception('TC1 should be smaller than TC2')
if TC1 + TC2 == 0:
if N == 1:
return 0
else:
return np.asarray([0] * N)
# define variables for trapezoidal distribution
A = TC1 * (1 - spread1)
B = TC2 * (1 + spread2)
if B < linf or A > lsup:
raise Exception('TC1, TC2, linf and lsup are not compatible')
c = (TC1 - A) / (B - A)
d = (TC2 - A) / (B - A)
loc = A
scale = B - A
dist = trapz.rvs(c, d, loc, scale, N)
truncdist = [i for i in dist if i >= linf]
truncdist = [i for i in truncdist if i <= lsup]
while len(truncdist) < N:
adddist = trapz.rvs(c, d, loc, scale, N - len(truncdist))
truncadddist = [i for i in adddist if i >= linf]
truncadddist = [i for i in truncadddist if i <= lsup]
truncdist = truncdist + truncadddist
if N == 1:
return truncdist[0]
else:
return np.asarray(truncdist)
def get_random_value(self):
"""
create a trapezoidal distribution based on the fuzzy number
and get a random value from it
"""
# trapezoidal distribution parameters
loc = self.value[0]
scale = self.value[3] - self.value[0]
c = (self.value[1] - self.value[0]) / scale
d = (self.value[2] - self.value[0]) / scale
# random value from trapezoidal distribution
r = trapz.rvs(c, d, loc=loc, scale=scale)
return r
# Display the probability density function (``pdf``): x = np.linspace(trapz.ppf(0.01, c, d), trapz.ppf(0.99, c, d), 100) ax.plot(x, trapz.pdf(x, c, d), 'r-', lw=5, alpha=0.6, label='trapz pdf') # Alternatively, the distribution object can be called (as a function) # to fix the shape, location and scale parameters. This returns a "frozen" # RV object holding the given parameters fixed. # Freeze the distribution and display the frozen ``pdf``: rv = trapz(c, d) ax.plot(x, rv.pdf(x), 'k-', lw=2, label='frozen pdf') # Check accuracy of ``cdf`` and ``ppf``: vals = trapz.ppf([0.001, 0.5, 0.999], c, d) np.allclose([0.001, 0.5, 0.999], trapz.cdf(vals, c, d)) # True # Generate random numbers: r = trapz.rvs(c, d, size=1000) # And compare the histogram: ax.hist(r, normed=True, histtype='stepfilled', alpha=0.2) ax.legend(loc='best', frameon=False) plt.show()
for method_i, method in enumerate(methods): #for each method
print "running {}".format(str(method))
for trial in tqdm(range(max_trials)): #for each possible trial run
trial_data[method_i].append([])
a_k = [random() for k in range(K)] #generate a bandit problem
b_k = [random() for k in range(K)]
for k in range(K):
if a_k[k] > b_k[k]:
a_k[k], b_k[k] = b_k[k], a_k[k]
means = [trapz.mean(a_k[i], b_k[i]) for i in range(K)]
max_mean = max(means)
samples = [[] for i in range(K)]
len_samples = [0 for i in range(K)]
for t in range(1, max_time + 1): #run the method
arm = method(samples, t)
samples[arm].append(trapz.rvs(a_k[arm], b_k[arm]))
len_samples[arm] += 1
trial_data[method_i][-1].append(
sum([len_samples[i] * (max_mean - means[i])
for i in range(K)]))
# transpose data arrays and calculate average regrets
rearranged_trial_data = [list(map(list, zip(*l))) for l in trial_data]
for i in range(len(rearranged_trial_data)):
for j in range(len(rearranged_trial_data[i])):
vv = rearranged_trial_data[i][j]
rearranged_trial_data[i][j] = sum(vv) * 1.0 / len(vv)
#output
with open("data22.csv", "w") as f:
for r in rearranged_trial_data:
def get_trapezoidal():
return trapz.rvs(0.2, 0.8, size=1)[0]