def current_percentile(percent):
if is_distributed:
# Flatten out the deque of lists and calculate the percentile to be returned
return percentile(
sorted([
item for sublist in master_response_times for item in sublist
]), percent)
else:
return percentile(sorted(master_response_times), percent)
def interquartile(x,type):
if(type=='acc'):
mean,percentile_75 = st.percentile(x,0.75,tol)
mean,percentile_25 = st.percentile(x,0.25,tol)
IQ = percentile_75-percentile_25
if(type=='braking'):
mean,percentile_75 = st.percentile(x,0.75,tol)
mean,percentile_25 = st.percentile(x,0.25,tol)
IQ = abs(percentile_75-percentile_25)
return IQ
def calc_smiley_face(val, vals):
sorted_vals = sorted([val] + vals)
lbound = stats.percentile(sorted_vals, 0.33333)
ubound = stats.percentile(sorted_vals, 0.66666)
if val >= ubound:
perc = "green"
elif val >= lbound:
perc = "yellow"
else:
perc = "red"
return perc
def GetSimulationPercentile(self, I, p, date=False):
variates = self.GetSimulationVariates(ID = I)
variate = stats.percentile(variates, p)
if date == True:
return self.GetNetworkStart(asobject = True) + datetime.timedelta(variate)
elif date == False:
return int(variate)
def sql_stats(meters, radius, tables):
cache_o = {}
to_query = {}
keys = []
lls = []
for m in meters:
p_center = [int(round(m[0])), int(round(m[1]))]
key = str(p_center[0]) + '_' + str(p_center[1])
if not key in to_query:
to_query[key] = True
keys.append(key)
lls.append(gm.MetersToLatLon(p_center[0], p_center[1]))
points = points_within_radius(gm.MetersToLatLon(m[0], m[1]), radius)[0]
for mp in points:
p = [int(round(mp[0])), int(round(mp[1]))]
key = str(p[0]) + '_' + str(p[1])
if not key in to_query:
to_query[key] = True
keys.append(key)
lls.append(gm.MetersToLatLon(p[0], p[1]))
offsets = get_offsets_linear(lls, tables)
for i in range(0,len(keys)):
cache_o[keys[i]] = offsets[i]
percentiles = []
for m in meters:
p_center = [int(round(m[0])), int(round(m[1]))]
key = str(p_center[0]) + '_' + str(p_center[1])
offset_center = cache_o[key]
points = points_within_radius(gm.MetersToLatLon(m[0], m[1]), radius)[0]
offsets = []
for mp in points:
p = [int(round(mp[0])), int(round(mp[1]))]
key = str(p[0]) + '_' + str(p[1])
offsets.append(cache_o[key])
percentiles.append(percentile(offsets, offset_center))
return percentiles
def calculate_stats_num(self, name, per = [5,25,50,75,95]):
# get columnt instance
Col = self.get_column(name)
# type validation
assert Col.type == 'numerical', 'only possible numerical columns.'
# get data
data = self.get_data(name)
# initialize
dstats = dict()
# calculate statistics
dstats['mean'] = stats.mean(data)
dstats['median'] = stats.median(data)
dstats['std'] = stats.std(data)
dstats['min'] = stats.min(data)
dstats['max'] = stats.max(data)
dstats['skew'] = stats.skew(data)
dstats['kurtosis'] = stats.kurtosis(data)
for ip in per:
dstats['per%s'%ip] = stats.percentile(data, ip)
# return
Col.stats = dstats
def current_percentile(percent):
if is_distributed:
# Flatten out the deque of lists and calculate the percentile to be returned
return percentile(sorted([item for sublist in response_times for item in sublist]), percent)
else:
return percentile(sorted(response_times), percent)
