def improvement(length):
"""
This function evaluates the ratio of the badness (defined
in this program as the height of the bar times the length
of the bar) of regularly blocked data to the badness of
dynamically blocked data on randomly generated data
of given length.
"rblocked" stands for "regularly blocked"
"dblocked" stands for "dynamically blocked"
"""
total_rblocked_badness = 0
total_dblocked_badness = 0
for i in range(N):
dat = [normal() + 2*sin(j / 25) for j in range(length)]
comp = compartmentalize(dat, max_length=MAX_BIN_LENGTH, max_pval=2.0/BIN_SIZE)
lengths_list = [[comp[0][i + 1][0] - comp[0][i][0]] * (comp[0][i + 1][0] - comp[0][i][0]) for i in range(len(comp[0]) - 1)]
average_length = average(list(itertools.chain(*lengths_list)))
total_dblocked_badness += sum([(comp[0][i][2] - comp[0][i][1]) * (comp[0][i + 1][0] - comp[0][i][0]) for i in range(len(comp[0]) - 1)])
rbadness = 0
for start in range(0, length, average_length):
if start + average_length > length:
start = start - average_length
else:
minimum = float("inf")
maximum = float("-inf")
for pos in range(start, start + average_length):
minimum = min(minimum, dat[pos])
maximum = max(maximum, dat[pos])
rbadness += (maximum - minimum) * average_length
total_rblocked_badness += rbadness * length / (start + average_length)
return total_rblocked_badness / total_dblocked_badness
def improvement(length):
"""
This function evaluates the ratio of the badness
of regularly blocked data to the badness of
dynamically blocked data on randomly generated data
of given length.
"rblocked" stands for "regularly blocked"
"dblocked" stands for "dynamically blocked"
"""
total_rblocked_badness = 0
total_dblocked_badness = 0
for i in range(N):
dat = [normal() + 2*sin(j / 25) for j in range(length)]
comp = compartmentalize(dat, max_length=MAX_BIN_LENGTH, max_pval=2.0/BIN_SIZE)
total_dblocked_badness += comp[1]
rbadness = 0
for start in range(0, length, BIN_SIZE):
minimum = float("inf")
maximum = -float("inf")
for pos in range(start, start + BIN_SIZE):
if dat[pos] < minimum:
minimum = dat[pos]
if dat[pos] > maximum:
maximum = dat[pos]
rbadness += (maximum - minimum) * barmap[BIN_SIZE]
total_rblocked_badness += rbadness
return total_rblocked_badness / total_dblocked_badness
