def generate_report(tuple_list, filepath, mode='variance'):
avg = []
for test in tuple_list:
testid, branchid, platformid = test[:3]
data_dict = compare.getGraphData(testid, branchid, platformid)
week_avgs = []
if data_dict:
data = data_dict['test_runs']
data.sort(key=lambda x: x[3])
data = data[int(0.1*len(data)):int(0.9*len(data) + 1)]
time_dict = collections.OrderedDict()
days = {}
for point in data:
time = datetime.fromtimestamp(point[2]).strftime('%Y-%m-%d')
time_dict[time] = time_dict.get(time, []) + [point[3]]
for time in time_dict:
runs = len(time_dict[time])
weekday = datetime.strptime(time, '%Y-%m-%d').strftime('%A')
variance = numpy.var(time_dict[time])
if mode == 'variance':
days[weekday] = days.get(weekday, []) + [variance]
elif mode == 'count':
days[weekday] = days.get(weekday, []) + [runs]
line = ["-".join(test[3:])]
for day in day_name:
if mode == 'variance':
# removing top and bottom 10% to reduce outlier influence
tenth = len(days[day])/10
average = numpy.average(
sorted(days[day])[tenth:tenth*9 + 1]
)
elif mode == 'count':
average = numpy.average(days[day])
line.append("%.3f" % average)
week_avgs.append(average)
outliers = is_normal(week_avgs)
for j in range(7):
if j in outliers:
line[j + 1] = "**" + str(line[j + 1]) + "**"
avg.append(line)
with open(filepath, 'wb') as report:
avgs_header = csv.writer(report, quoting=csv.QUOTE_ALL)
avgs_header.writerow(['test-platform'] + list(day_name))
for line in avg:
out = csv.writer(report, quoting=csv.QUOTE_ALL)
out.writerow(line)
def generate_report(tuple_list, filepath, mode="variance"):
avg = []
for test in tuple_list:
testid, branchid, platformid = test[:3]
data_dict = compare.getGraphData(testid, branchid, platformid)
week_avgs = []
if data_dict:
data = data_dict["test_runs"]
data.sort(key=lambda x: x[3])
data = data[int(0.1 * len(data)):int(0.9 * len(data) + 1)]
time_dict = collections.OrderedDict()
days = {}
for point in data:
time = datetime.fromtimestamp(point[2]).strftime("%Y-%m-%d")
time_dict[time] = time_dict.get(time, []) + [point[3]]
for time in time_dict:
runs = len(time_dict[time])
weekday = datetime.strptime(time, "%Y-%m-%d").strftime("%A")
variance = numpy.var(time_dict[time])
if mode == "variance":
days[weekday] = days.get(weekday, []) + [variance]
elif mode == "count":
days[weekday] = days.get(weekday, []) + [runs]
line = ["-".join(test[3:])]
for day in day_name:
if mode == "variance":
# removing top and bottom 10% to reduce outlier influence
# pylint --py3k W1619
tenth = len(days[day]) / 10
average = numpy.average(
sorted(days[day])[tenth:tenth * 9 + 1])
elif mode == "count":
average = numpy.average(days[day])
line.append("%.3f" % average)
week_avgs.append(average)
outliers = is_normal(week_avgs)
for j in six.moves.range(7):
if j in outliers:
line[j + 1] = "**" + str(line[j + 1]) + "**"
avg.append(line)
with open(filepath, "wb") as report:
avgs_header = csv.writer(report, quoting=csv.QUOTE_ALL)
avgs_header.writerow(["test-platform"] + list(day_name))
for line in avg:
out = csv.writer(report, quoting=csv.QUOTE_ALL)
out.writerow(line)