def testVar(self):
'''Calculate the biased variance of a series'''
ts = dynts.timeseries(date=datepopulate(10),
data=range(1, 11),
backend=self.backend)
self.assertAlmostEqual(ts.var()[0], 8.25, places)
self.assertAlmostEqual(ts.var(ddof=1)[0], 9.166667, places)
def testHashEmpty(self):
ts = timeseries(dtype=object)
h = ts.ashash()
self.assertFalse(h)
dt1 = date.today()
dt0 = date.today() - timedelta(days=1)
h[dt1] = [56, 48]
h[dt0] = 'this is a string'
self.assertEqual(len(h), 2)
ts = h.getts()
self.assertEqual(len(ts), 2)
self.assertEqual(ts[0][0], 'this is a string')
self.assertEqual(ts[1][0], [56, 48])
def testHashEmpty(self):
ts = timeseries(dtype = object)
h = ts.ashash()
self.assertFalse(h)
dt1 = date.today()
dt0 = date.today()-timedelta(days=1)
h[dt1] = [56,48]
h[dt0] = 'this is a string'
self.assertEqual(len(h),2)
ts = h.getts()
self.assertEqual(len(ts),2)
self.assertEqual(ts[0][0], 'this is a string')
self.assertEqual(ts[1][0], [56,48])
def _unwind(self, values, backend, **kwargs):
sdata = values[self.value]
if istimeseries(sdata):
return sdata
else:
ts = timeseries(name=str(self),
date=sdata['date'],
data=sdata['value'],
backend=backend)
# Uses this hack to make sure timeseries are ordered
# Lots of room for performance improvement
hash = ts.ashash()
hash.modified = True
values[ts.name] = hash.getts()
return ts
def _unwind(self, values, backend, **kwargs):
sdata = values[self.value]
if istimeseries(sdata):
return sdata
else:
ts = timeseries(name = str(self),
date = sdata['date'],
data = sdata['value'],
backend = backend)
# Uses this hack to make sure timeseries are ordered
# Lots of room for performance improvement
hash = ts.ashash()
hash.modified = True
ts = hash.getts()
values[ts.name] = ts
return ts
def load_data(self, result):
loads = self.pickler.loads
vloads = self.value_pickler.loads
dt, va = result
if result[0] and va:
dates = ny.array([loads(t) for t in dt])
fields = []
vals = []
if not isinstance(va, Mapping):
va = dict(va)
for f in sorted(va):
fields.append(f)
data = va[f]
vals.append((vloads(v) for v in data))
values = ny.array(list(zip(*vals)))
name = tsname(*fields)
else:
name = None
dates = None
values = None
return timeseries(name=name, date=dates, data=values)
def load_data(self, result):
loads = self.pickler.loads
vloads = self.value_pickler.loads
dt, va = result
if result[0] and va:
dates = ny.array([loads(t) for t in dt])
fields = []
vals = []
if not isinstance(va, Mapping):
va = dict(va)
for f in sorted(va):
fields.append(f)
data = va[f]
vals.append((vloads(v) for v in data))
values = ny.array(list(zip(*vals)))
name = tsname(*fields)
else:
name = None
dates = None
values = None
return timeseries(name=name, date=dates, data=values)
def testEmptyHashWrapper(self):
ts = timeseries()
hash = ts.ashash()
self.assertFalse(hash)
def testEmptyHashWrapper(self):
ts = timeseries()
hash = ts.ashash()
self.assertFalse(hash)
def timeseries(self, name='', date=None, data=None):
return timeseries(name=name, date=date, data=data,
backend=self.backend)
def timeseries(self, name='', date=None, data=None):
return timeseries(name=name,
date=date,
data=data,
backend=self.backend)
def randomts(size = 100, cols = 1, start = None, delta = 1,
generator = None, backend=None, name='randomts'):
from dynts import timeseries
dates = datepopulate(size,start=start,delta=delta)
data = populate(size,cols=cols,generator=generator)
return timeseries(name=name,backend=backend,date=dates,data=data)
def testVar(self):
'''Calculate the biased variance of a series'''
ts = timeseries(date = datepopulate(10), data = range(1,11), backend = self.backend)
self.assertAlmostEqual(ts.var()[0],8.25)
self.assertAlmostEqual(ts.var(ddof=1)[0],9.166667)
def get(self):
ts = timeseries(dtype=object)
self.assertTrue(ts.is_object)
self.assertEqual(ts.dtype, np.dtype(object))
def get(self):
ts = timeseries(dtype = object)
self.assertTrue(ts.is_object)
self.assertEqual(ts.dtype, np.dtype(object))
