def check(samples_size, alpha, beta, eps, delta, parameter):
data = [int(i) for i in normal(0, parameter, samples_size)]
m = min(data)
data = [i - m for i in data]
max_sample = max(data)
dim = ceil(log2(max_sample + 1))
end_domain = 2**int(dim)
try:
san = sanitize(data, (0, end_domain), alpha, beta, eps, delta)
result = cdf_comp(san, data)
if result == 0:
plot_san_and_original(data, san)
except ValueError:
result = -1
return result
def check(samples_size, alpha, beta, eps, delta, parameter):
data = [int(i) for i in normal(0, parameter, samples_size)]
m = min(data)
data = [i-m for i in data]
max_sample = max(data)
dim = ceil(log2(max_sample + 1))
end_domain = 2**int(dim)
try:
san = sanitize(data, (0, end_domain), alpha, beta, eps, delta)
result = cdf_comp(san, data)
if result == 0:
plot_san_and_original(data, san)
except ValueError:
result = -1
return result
f2 = ECDF(data2)
return 1-sum(1 for c in xrange(m) if abs(f1(c) - f2(c)) <= alpha)/m
a, b, e, d = 0.1, 0.1, 0.5, 2**-20
b *= a / 231
samples_no = 5000
parameter = 5
data = [int(i) for i in normal(0, parameter, samples_no)]
m = min(data)
data = [i-m for i in data]
print len(data)
max_sample = max(data)
print max_sample
dim = ceil(log2(max_sample + 1))
print dim
end_domain = 2**int(dim)
start_time = time.time()
san = sanitize(data, (0, end_domain), a, b, e, d)
run_time = time.time() - start_time
print max(san)
print cdf_comp(san, data, a)
sorted_san = sorted(san)
i_max_san = searchsorted(sorted_san, max_sample)
limited_san = sorted_san[:i_max_san]
plot_cdf(data)
plot_cdf(limited_san)
plt.show()
print "run-time: %.2f seconds" % run_time
