def test_fit_remove_add_trend(order, n_instances, n_timepoints):
coefs = np.random.normal(size=order + 1).reshape(-1, 1)
x = np.column_stack(
[
_generate_polynomial_series(n_timepoints, order, coefs=coefs)
for _ in range(n_instances)
]
).T
# assert x.shape == (n_samples, n_obs)
# check shape of fitted coefficients
coefs = fit_trend(x, order=order)
assert coefs.shape == (n_instances, order + 1)
def test_fit_remove_add_trend(order, n_samples, n_obs):
# generate random polynomial series data
coefs = np.random.normal(size=order + 1).reshape(-1, 1)
x = np.column_stack([
generate_polynomial_series(n_obs, order, coefs=coefs)
for _ in range(n_samples)
]).T
# assert x.shape == (n_samples, n_obs)
# check shape of fitted coefficients
coefs = fit_trend(x, order=order)
assert coefs.shape == (n_samples, order + 1)
# test if trend if properly remove when given true order
xt = remove_trend(x, coefs)
np.testing.assert_array_almost_equal(xt, np.zeros(x.shape))
# test inverse transform restores original series
xit = add_trend(xt, coefs=coefs)
np.testing.assert_array_almost_equal(x, xit)
def transform(self, X, y=None):
"""Transform X.
Parameters
----------
X : nested pandas DataFrame of shape [n_samples, n_features]
Nested dataframe with time-series in cells.
Returns
-------
Xt : pandas DataFrame
Transformed pandas DataFrame with same number of rows and one column for each generated interval.
"""
if self.check_input:
if not isinstance(X, pd.DataFrame):
raise ValueError(f"Input must be pandas DataFrame, but found: {type(X)}")
if X.shape[1] > 1:
raise NotImplementedError(f"Currently does not work on multiple columns")
self._input_shape = X.shape
# keep time index as trend depends on it, e.g. to carry forward trend in inverse_transform
self._time_index = get_time_index(X)
# convert into tabular format
tabulariser = Tabulariser()
Xs = tabulariser.transform(X.iloc[:, :1])
# fit polynomial trend
self.coefs_ = fit_trend(Xs, order=self.order)
# remove trend
Xt = remove_trend(Xs, coefs=self.coefs_, time_index=self._time_index)
# convert back into nested format
Xt = tabulariser.inverse_transform(pd.DataFrame(Xt))
Xt.columns = X.columns
return Xt
def _compute_trend(y):
# Trend calculated through least squares regression.
coefs = fit_trend(y.values.reshape(1, -1), order=1)
return coefs[0, 0] / 2