def test_accumulate_twice(padding):
#If you accumulate twice, it should be the same as delta1*delta2
#This is unimplemented
ts1 = TimeSeries(sample_data)
ts2 = TimeSeries(sample_data)
ts1.accumulate(10,padding=padding)
ts2.accumulate(2,padding=padding)
with pytest.raises(NotImplementedError):
ts2.accumulate(5,padding=padding)
def test_get_y():
ts = TimeSeries(sample_data)
y = ts.get_y()
#Every value should be equal to the to the first column in ts.data
for i in range(len(y)):
assert(y[i]==ts.data[i][1])
def test_init__correct_sum():
ts = TimeSeries(sample_data)
#The total should be the sum of all the individual posts in sample_data
sum1 = 0
for data in ts.data:
sum1 += data[1]
assert(sum1==200+50+200+400-300+100)
def test_accumulate__steps(padding, sample_key):
ts = TimeSeries(sample_data[sample_key])
ts.accumulate(1, "Month", padding=padding)
for i in range(1, len(ts.data)):
assert ((ts.data[i][0] -
ts.data[i - 1][0]).days == calendar.monthrange(
ts.data[i - 1][0].year, ts.data[i - 1][0].month)[1])
def test_accumulate__end_date(padding, sample_key):
sd = sample_data[sample_key]
ts = TimeSeries(sd)
ts.accumulate(1, "Year", padding=padding)
assert (ts.get_x()[-1] == expected_end_date(sd[-1][0], padding))
assert (ts.get_x()[-1].day == 1)
assert (ts.get_x()[-1].month == 1)
def test_init__no_missing_dates():
ts = TimeSeries(sample_data)
#All dates should be present between the first and the last, an only occur once
last_date = ts.data[0][0]
for data in ts.data[1:]:
assert(last_date+datetime.timedelta(1)==data[0])
last_date = data[0]
def test_accumulate__steps(padding, sample_key):
ts = TimeSeries(sample_data[sample_key])
ts.accumulate(1, "Year", padding=padding)
for i in range(1, len(ts.data)):
assert (ts.data[i - 1][0].day == 1)
assert (ts.data[i - 1][0].month == 1)
assert (ts.data[i][0].year - ts.data[i - 1][0].year == 1)
def test_accumulate__len(padding, sample_key):
ts = TimeSeries(sample_data[sample_key])
original_len = len(ts.data)
start_date = expected_start_date(ts.data[0][0], padding)
end_date = expected_end_date(ts.data[-1][0], padding)
expected_len = ((end_date - start_date).days + 7) / 7
ts.accumulate(1, "Week", padding=padding)
assert (len(ts.data) == expected_len)
def test_accumulate__len(padding,sample_key,delta):
ts = TimeSeries(sample_data[sample_key])
original_len = len(ts.data)
start_date = expected_start_date(ts.data[0][0],delta,original_len,padding)
end_date = expected_end_date(ts.data[-1][0],delta,original_len,padding)
expected_len = (end_date-start_date).days/delta+1
ts.accumulate(delta,"Day",padding=padding)
assert(len(ts.data) == expected_len)
def test_accumulate__len(padding, sample_key):
ts = TimeSeries(sample_data[sample_key])
original_len = len(ts.data)
start_date = expected_start_date(ts.data[0][0], padding)
end_date = expected_end_date(ts.data[-1][0], padding)
expected_len = end_date.year - start_date.year + 1
ts.accumulate(1, "Year", padding=padding)
assert (len(ts.data) == expected_len)
def test_accumulate__sum(padding,sample_key,delta):
sd = sample_data[sample_key]
ts = TimeSeries(sd)
ts.accumulate(delta,"Day",padding=padding)
for i in range(len(ts.data)):
cum_sum = 0
for data in sd:
if ts.data[i][0]<=data[0]<ts.data[i][0]+datetime.timedelta(delta):
cum_sum+=data[1]
assert(ts.data[i][1]==cum_sum)
def test_accumulate__sum(padding, sample_key):
sd = sample_data[sample_key]
ts = TimeSeries(sd)
ts.accumulate(1, "Month", padding=padding)
for i in range(len(ts.data)):
cum_sum = 0
for data in sd:
if ts.data[i][0] <= data[0] < ts.data[i][0] + datetime.timedelta(
calendar.monthrange(ts.data[i][0].year,
ts.data[i][0].month)[1]):
cum_sum += data[1]
assert (ts.data[i][1] == cum_sum)
def test_accumulate__sum(padding, sample_key):
#If padding false, the first day should be skipped (in this case)
#If padding is true, only the first day should be summed (together with two empty days)
sd = sample_data[sample_key]
ts = TimeSeries(sd)
ts.accumulate(1, "Week", padding=padding)
for i in range(len(ts.data)):
cum_sum = 0
for data in sd:
if ts.data[i][0] <= data[0] < ts.data[i][0] + datetime.timedelta(
7):
cum_sum += data[1]
assert (ts.data[i][1] == cum_sum)
def test_accumulate__start_date(padding, sample_key):
sd = sample_data[sample_key]
ts = TimeSeries(sd)
ts.accumulate(1, "Month", padding=padding)
assert (ts.get_x()[0] == expected_start_date(sd[0][0], padding))
assert (ts.get_x()[0].day == 1)
def test_accumulate__empty():
#It should be possible to initialize with an empty list
ts = TimeSeries([])
ts.accumulate(2)
assert(ts.data==[])
def test_init__empty():
#It should be possible to initialize with an empty list
ts = TimeSeries([])
assert(ts.data==[])
def test_accumulate__end_date(padding,sample_key,delta):
sd = sample_data[sample_key]
ts = TimeSeries(sd)
original_len = len(ts.data)
ts.accumulate(delta,"Day",padding=padding)
assert(ts.get_x()[-1]==expected_end_date(sd[-1][0],delta,original_len,padding))
def test_moving_average__empty():
#It should be possible to initialize with an empty list
ts = TimeSeries([])
ts.moving_average(3)
assert(ts.data==[])
def test_moving_average__first_val():
ts = TimeSeries(sample_data)
original_len = len(ts.data)
ts.moving_average(3)
assert(ts.data[0][-1]==(200+50)/3)
def test_moving_average__last_day(window):
ts = TimeSeries(sample_data)
original_len = len(ts.data)
ts.moving_average(window)
assert(ts.data[-1][0]==sample_data[-1][0])
def test_moving_average__first_day(window):
ts = TimeSeries(sample_data)
ts.moving_average(window)
assert(ts.data[0][0]==sample_data[0][0]+datetime.timedelta(window-1))
def test_accumulate__end_date(padding, sample_key):
sd = sample_data[sample_key]
ts = TimeSeries(sd)
ts.accumulate(1, "Week", padding=padding)
assert (ts.get_x()[-1] == expected_end_date(sd[-1][0], padding))
assert (ts.get_x()[-1].weekday() == 0)
def test_accumulate__steps(padding, sample_key):
ts = TimeSeries(sample_data[sample_key])
ts.accumulate(1, "Week", padding=padding)
for i in range(1, len(ts.data)):
assert ((ts.data[i][0] - ts.data[i - 1][0]).days == 7)
def test_moving_average__len(window):
ts = TimeSeries(sample_data)
original_len = len(ts.data)
ts.moving_average(window)
assert(len(ts.data)==original_len-window+1)
def get_timeseries(self):
return TimeSeries(self.get_data(), daterange=self._f_daterange)
def test_init__length():
ts = TimeSeries(sample_data)
#The lenght of the timeseries should be the number of number of days between first and last day of data 1
assert(len(ts.data)==11)