TSCFiniteDifference#
- class datafold.dynfold.TSCFiniteDifference(*, spacing='dt', scheme='center', diff_order=1, accuracy=2)[source]#
Bases:
BaseEstimator
,TSCTransformerMixin
Compute time derivative with finite difference scheme.
Note
The class internally uses the Python package findiff, which currently is optional in datafold. The class raises an ImportError if findiff is not installed.
- Parameters:
spacing (Union[str, float]) – The time difference between samples. If “dt” (str) then the time sampling frequency of a
TSCDataFrame.delta_time()
is used during fit.scheme (
Literal
['backward'
,'center'
,'forward'
]) – The finite difference scheme to apply, “center”, “backward” or “forward”.diff_order (
int
) – The derivative order.accuracy (
int
) – The convergence order of the finite difference scheme.
- Variables:
spacing – The resolved time difference between samples. Equals the parameter input if it was of type :class`float`.
See also
Methods Summary
fit
(X[, y])Set and validate time spacing between samples.
get_feature_names_out
([input_features])partial_fit
(X[, y])- rtype:
transform
(X)Compute the finite difference values.
Methods Documentation
- fit(X, y=None, **fit_params)[source]#
Set and validate time spacing between samples.
- Parameters:
X (TSCDataFrame, pandas.DataFrame, numpy.ndarray) – Data of shape (n_samples, n_features).
y (None) – ignored
- Returns:
self
- Return type:
- Raises:
TSCException – If time series data has not a constant time delta or the input X has not the same value as specified in spacing during initialization.
- transform(X)[source]#
Compute the finite difference values.
- Parameters:
X (TSCDataFrame, pandas.DataFrame, numpy.ndarray) – Data of shape (n_samples, n_features).
- Returns:
Transformed data of same shape and type as X.
- Return type:
- Raises:
TSCException – If input X has a different time delta than data during fit.