def test_normalization_bypercentile(self):
rdd = self.sc.parallelize([(0, array([1, 2, 3, 4, 5], dtype='float16'))])
data = TimeSeries(rdd, dtype='float16')
out = data.normalize('percentile', perc=20)
vals = out.first()[1]
assert_equals('float16', str(vals.dtype))
assert(allclose(vals, array([-0.42105, 0.10526, 0.63157, 1.15789, 1.68421]), atol=1e-3))
def test_normalization_bywindow(self):
y = array([1, 2, 3, 4, 5], dtype='float16')
rdd = self.sc.parallelize([(0, y)])
data = TimeSeries(rdd, dtype='float16')
out = data.normalize('window', window=2)
vals = out.first()[1]
assert_equals('float64', str(vals.dtype))
b_true = array([1.2, 1.4, 2.4, 3.4, 4.2])
result_true = (y - b_true) / (b_true + 0.1)
assert(allclose(vals, result_true, atol=1e-3))
out = data.normalize('window', window=6)
vals = out.first()[1]
b_true = array([1.6, 1.8, 1.8, 1.8, 2.6])
result_true = (y - b_true) / (b_true + 0.1)
assert(allclose(vals, result_true, atol=1e-3))
out = data.normalize('window-fast', window=2)
vals = out.first()[1]
assert_equals('float64', str(vals.dtype))
b_true = array([1, 1, 2, 3, 4])
result_true = (y - b_true) / (b_true + 0.1)
assert(allclose(vals, result_true, atol=1e-3))
out = data.normalize('window-fast', window=5)
vals = out.first()[1]
b_true = array([1, 1, 2, 3, 4])
result_true = (y - b_true) / (b_true + 0.1)
assert(allclose(vals, result_true, atol=1e-3))
def test_convolve(self):
dataLocal = array([1, 2, 3, 4, 5])
sig = array([1, 2, 3])
rdd = self.sc.parallelize([(0, dataLocal)])
data = TimeSeries(rdd)
betas = data.convolve(sig, mode='same')
assert(allclose(betas.values().collect()[0], array([4, 10, 16, 22, 22])))
def test_normalization_bymean(self):
rdd = self.sc.parallelize([(0, array([1, 2, 3, 4, 5], dtype='float16'))])
data = TimeSeries(rdd, dtype='float16')
out = data.normalize('mean')
vals = out.first()[1]
assert_equals('float16', str(vals.dtype))
assert(allclose(out.first()[1],
array([-0.64516, -0.32258, 0.0, 0.32258, 0.64516]), atol=1e-3))
def test_fourier(self):
dataLocal = [
array([1.0, 2.0, -4.0, 5.0, 8.0, 3.0, 4.1, 0.9, 2.3]),
array([2.0, 2.0, -4.0, 5.0, 3.1, 4.5, 8.2, 8.1, 9.1]),
]
rdd = self.sc.parallelize(zip(range(1, 3), dataLocal))
data = TimeSeries(rdd)
vals = data.fourier(freq=2)
assert(allclose(vals.select('coherence').values().collect()[0], 0.578664))
assert(allclose(vals.select('phase').values().collect()[0], 4.102501))
def test_crossCorr(self):
dataLocal = array([
array([1.0, 2.0, -4.0, 5.0, 8.0, 3.0, 4.1, 0.9, 2.3]),
array([2.0, 2.0, -4.0, 5.0, 3.1, 4.5, 8.2, 8.1, 9.1]),
])
sig = array([1.5, 2.1, -4.2, 5.6, 8.1, 3.9, 4.2, 0.3, 2.1])
rdd = self.sc.parallelize(zip(range(1, 3), dataLocal))
data = TimeSeries(rdd)
betas = data.crossCorr(signal=sig, lag=0)
assert(allclose(betas.values().collect()[0], corrcoef(dataLocal[0, :], sig)[0, 1]))
assert(allclose(betas.values().collect()[1], corrcoef(dataLocal[1, :], sig)[0, 1]))
betas = data.crossCorr(signal=sig, lag=2)
tol = 1E-5 # to handle rounding errors
assert(allclose(betas.values().collect()[0], array([-0.18511, 0.03817, 0.99221, 0.06567, -0.25750]), atol=tol))
assert(allclose(betas.values().collect()[1], array([-0.35119, -0.14190, 0.44777, -0.00408, 0.45435]), atol=tol))
def test_meanByWindow(self):
rdd = self.sc.parallelize([((0,), array([0, 1, 2, 3, 4, 5, 6], dtype='float16'))])
data = TimeSeries(rdd, dtype='float16')
assert(data.count() == 1)
test1 = data.meanByWindow(indices=[3, 5], window=2).first()[1]
assert(allclose(test1, [3, 4]))
test2 = data.meanByWindow(indices=[3, 5], window=3).first()[1]
assert(allclose(test2, [3, 4, 5]))
test3 = data.meanByWindow(indices=[3, 5], window=4).first()[1]
assert(allclose(test3, [2, 3, 4, 5]))
test4 = data.meanByWindow(indices=[3], window=4).first()[1]
assert(allclose(test4, [1, 2, 3, 4]))
def test_groupByWindow(self):
rdd = self.sc.parallelize([((0,), array([0, 1, 2, 3, 4, 5, 6], dtype='float16'))])
data = TimeSeries(rdd, dtype='float16')
test1 = data.groupByWindow(indices=[3, 5], window=2).values().collect()
assert(allclose(test1, [[2, 3], [4, 5]]))
test2 = data.groupByWindow(indices=[3, 5], window=3).values().collect()
assert(allclose(test2, [[2, 3, 4], [4, 5, 6]]))
test3 = data.groupByWindow(indices=[3, 5], window=4).values().collect()
assert(allclose(test3, [[1, 2, 3, 4], [3, 4, 5, 6]]))
test4 = data.groupByWindow(indices=[3, 4, 5], window=2).values().collect()
assert(allclose(test4, [[2, 3], [3, 4], [4, 5]]))
test5 = data.groupByWindow(indices=[3], window=2).values().collect()
assert(allclose(test5, [[2, 3]]))
def test_detrend(self):
rdd = self.sc.parallelize([(0, array([1, 2, 3, 4, 5]))])
data = TimeSeries(rdd).detrend('linear')
# detrending linearly increasing data should yield all 0s
assert(allclose(data.first()[1], array([0, 0, 0, 0, 0])))
