def test_timecorr():
data_dl = hyp.tools.format_data(data_list)
data_pdf = hyp.tools.format_data(pandas_dataframe)
data_npa = hyp.tools.format_data(numpy_array)
# data_npl = hyp.tools.format_data(numpy_array_list)
# data_rand = hyp.tools.format_data(random_numbers)
# these are now lists
assert isinstance(data_dl, list)
Test_dl = data_dl[0].shape[0]
Test_pdf = data_pdf[0].shape[0]
Test_npa = data_npa[0].shape[0]
#Test returns the shape of the weights_function
# Test_npl= data_npl[0].shape[0]
# Test_rand= data_rn[0].shape[0]
assert isinstance(Test_pdf, int)
dl_tester = gaussian_weights(Test_dl, params=gaussian_params)
pdf_tester = gaussian_weights(Test_pdf, params=gaussian_params)
npa_tester = gaussian_weights(Test_npa, params=gaussian_params)
# thrid_tester = gaussian_weights(T3, params=gaussian_params)
# fourth_tester = gaussian_weights(T4, params=gaussian_params)
assert isinstance(npa_tester, np.ndarray)
def test_wisfc():
weights = gaussian_weights(T, params=gaussian_params)
w_list = wisfc(data, weights)
assert isinstance(w_list, list)
w_array = wisfc(template_data, weights)
assert isinstance(w_array, np.ndarray)
def test_wcorr():
weights = gaussian_weights(T, params=gaussian_params)
corrs = wcorr(template_data[:, 0][:, np.newaxis],
template_data[:, 1][:, np.newaxis], weights)
col_1 = np.atleast_2d(data_sim[:, 0]).T
col_2 = np.atleast_2d(data_sim[:, 1]).T
corrs_col_arrays = np.squeeze(wcorr(data_sim, data_sim, weights_sim))
corrs_multidim = wcorr(col_1, data_sim, weights_sim)
corrs_col = np.squeeze(wcorr(col_1, col_1, weights_sim))
corrs_col_neg = np.squeeze(wcorr(-col_1, col_1, weights_sim))
corrs_col_12 = np.squeeze(wcorr(col_2, col_1, weights_sim))
# correlate one column with itself and the same column in larger array is the same
assert (np.allclose(corrs_col, corrs_multidim[0][0]))
# correlating a timeseries with -1 times itself produces -1's
assert corrs_col.mean() == 1
# correlate one column with the second column and the same column in larger array is the same
assert (np.allclose(corrs_col_12, corrs_multidim[0][1]))
# correlating a timeseries with -1 times itself produces negative correlations
assert (np.allclose(-corrs_col, corrs_col_neg))
# correlating a timeseries with -1 times itself produces -1's
assert corrs_col_neg.mean() == -1
# check if corresponding columns in 3d array produces 1
assert (np.isclose(corrs_col_arrays[4, 4, 500], 1))
# check if toeplitz matrix is produced
assert (np.allclose(corrs_col_arrays[:, :, 500], R, atol=.2))
# check if corrs is a numpy array
assert isinstance(corrs, np.ndarray)
def test_gaussian_weights():
test_gw = gaussian_weights(T, params=gaussian_params)
assert isinstance(test_gw, np.ndarray)
'params': {
'scale': width
}
}
try_data = []
repdata = 4
for i in range(repdata):
try_data.append(data_sim)
try_data = np.array(try_data)
gps = {'var': 100}
T_sim = data_sim.shape[0]
weights_sim = gaussian_weights(T_sim, gps)
template_data = np.cumsum(np.random.randn(T, D), axis=0)
data = []
for s in np.arange(S):
data.append(template_data + np.multiply(
0.1, np.random.randn(template_data.shape[0], template_data.shape[1])))
def test_gaussian_weights():
test_gw = gaussian_weights(T, params=gaussian_params)
assert isinstance(test_gw, np.ndarray)