BayesianBasisExpansionTimeSeries

class causalpy.pymc_models.BayesianBasisExpansionTimeSeries

Bayesian Structural Time Series Model.

This model allows for the inclusion of trend, seasonality (via Fourier series), and optional exogenous regressors.

\[\begin{split}\text{trend} &\sim \text{LinearTrend}(...) \\ \text{seasonality} &\sim \text{YearlyFourier}(...) \\ \beta &\sim \mathrm{Normal}(0, \sigma_{\beta}) \quad \text{(if X is provided)} \\ \sigma &\sim \mathrm{HalfNormal}(\sigma_{err}) \\ \mu &= \text{trend_component} + \text{seasonality_component} + X \cdot \beta \quad \text{(if X is provided)} \\ y &\sim \mathrm{Normal}(\mu, \sigma)\end{split}\]

Parameters:

• n_order (int, optional) – The number of Fourier components for the yearly seasonality. Defaults to 3. Only used if seasonality_component is None.

• n_changepoints_trend (int, optional) – The number of changepoints for the linear trend component. Defaults to 10. Only used if trend_component is None.

• prior_sigma (float, optional) – Prior standard deviation for the observation noise. Defaults to 5.

• trend_component (Optional[Any], optional) – A custom trend component model. If None, the default pymc-marketing LinearTrend component is used. Must have an apply(time_data) method that returns a PyMC tensor.

• seasonality_component (Optional[Any], optional) – A custom seasonality component model. If None, the default pymc-marketing YearlyFourier component is used. Must have an apply(time_data) method that returns a PyMC tensor.

• sample_kwargs (dict, optional) – A dictionary of kwargs that get unpacked and passed to the pymc.sample() function. Defaults to an empty dictionary.

Methods

BayesianBasisExpansionTimeSeries.__init__([...])
BayesianBasisExpansionTimeSeries.add_coord(name)	Register a dimension coordinate with the model.
BayesianBasisExpansionTimeSeries.add_coords(...)	Vectorized version of Model.add_coord.
BayesianBasisExpansionTimeSeries.add_named_variable(var)	Add a random graph variable to the named variables of the model.
BayesianBasisExpansionTimeSeries.build_model(X, ...)	Defines the PyMC model.
BayesianBasisExpansionTimeSeries.calculate_cumulative_impact(impact)
BayesianBasisExpansionTimeSeries.calculate_impact(...)
BayesianBasisExpansionTimeSeries.check_start_vals(...)	Check that the logp is defined and finite at the starting point.
BayesianBasisExpansionTimeSeries.compile_d2logp([...])	Compiled log probability density hessian function.
BayesianBasisExpansionTimeSeries.compile_dlogp([...])	Compiled log probability density gradient function.
BayesianBasisExpansionTimeSeries.compile_fn(outs, *)	Compiles a PyTensor function.
BayesianBasisExpansionTimeSeries.compile_logp([...])	Compiled log probability density function.
BayesianBasisExpansionTimeSeries.copy()	Clone the model.
BayesianBasisExpansionTimeSeries.create_value_var(...)	Create a TensorVariable that will be used as the random variable's "value" in log-likelihood graphs.
BayesianBasisExpansionTimeSeries.d2logp([...])	Hessian of the models log-probability w.r.t.
BayesianBasisExpansionTimeSeries.debug([...])	Debug model function at point.
BayesianBasisExpansionTimeSeries.dlogp([...])	Gradient of the models log-probability w.r.t.
BayesianBasisExpansionTimeSeries.eval_rv_shapes()	Evaluate shapes of untransformed AND transformed free variables.
BayesianBasisExpansionTimeSeries.fit(X, y, ...)	Draw samples from posterior, prior predictive, and posterior predictive distributions, placing them in the model's idata attribute. :type X: Optional[ndarray] :param X: NumPy array of exogenous regressors. Can be None or an array with 0 columns if no exogenous variables. :type X: np.ndarray or None :type y: ndarray :param y: The target variable. :type y: np.ndarray :type coords: Dict[str, Any] :param coords: Coordinates dictionary. Must contain "datetime_index" (pd.DatetimeIndex). If X is provided and has columns, coords must also contain "coeffs" (List[str]). :type coords: dict.
BayesianBasisExpansionTimeSeries.get_context([...])
BayesianBasisExpansionTimeSeries.initial_point([...])	Compute the initial point of the model.
BayesianBasisExpansionTimeSeries.logp([...])	Elemwise log-probability of the model.
BayesianBasisExpansionTimeSeries.logp_dlogp_function([...])	Compile a PyTensor function that computes logp and gradient.
BayesianBasisExpansionTimeSeries.make_obs_var(...)	Create a TensorVariable for an observed random variable.
BayesianBasisExpansionTimeSeries.name_for(name)	Check if name has prefix and adds if needed.
BayesianBasisExpansionTimeSeries.name_of(name)	Check if name has prefix and deletes if needed.
BayesianBasisExpansionTimeSeries.point_logps([...])	Compute the log probability of point for all random variables in the model.
BayesianBasisExpansionTimeSeries.predict(X, ...)	Predict data given input X and coords for prediction period.
BayesianBasisExpansionTimeSeries.print_coefficients(labels)
BayesianBasisExpansionTimeSeries.profile(outs, *)	Compile and profile a PyTensor function which returns outs and takes values of model vars as a dict as an argument.
BayesianBasisExpansionTimeSeries.register_data_var(data)	Register a data variable with the model.
BayesianBasisExpansionTimeSeries.register_rv(...)	Register an (un)observed random variable with the model.
BayesianBasisExpansionTimeSeries.replace_rvs_by_values(...)	Clone and replace random variables in graphs with their value variables.
BayesianBasisExpansionTimeSeries.score(X, y, ...)	Score the Bayesian R2.
BayesianBasisExpansionTimeSeries.set_data(...)	Change the values of a data variable in the model.
BayesianBasisExpansionTimeSeries.set_dim(...)	Update a mutable dimension.
BayesianBasisExpansionTimeSeries.set_initval(...)	Set an initial value (strategy) for a random variable.
BayesianBasisExpansionTimeSeries.shape_from_dims(dims)
BayesianBasisExpansionTimeSeries.to_graphviz(*)	Produce a graphviz Digraph from a PyMC model.

Attributes

basic_RVs	List of random variables the model is defined in terms of.
continuous_value_vars	All the continuous value variables in the model.
coords	Coordinate values for model dimensions.
datalogp	PyTensor scalar of log-probability of the observed variables and potential terms.
dim_lengths	The symbolic lengths of dimensions in the model.
discrete_value_vars	All the discrete value variables in the model.
isroot
observedlogp	PyTensor scalar of log-probability of the observed variables.
parent
potentiallogp	PyTensor scalar of log-probability of the Potential terms.
prefix
root
unobserved_RVs	List of all random variables, including deterministic ones.
unobserved_value_vars	List of all random variables (including untransformed projections), as well as deterministics used as inputs and outputs of the model's log-likelihood graph.
value_vars	List of unobserved random variables used as inputs to the model's log-likelihood (which excludes deterministics).
varlogp	PyTensor scalar of log-probability of the unobserved random variables (excluding deterministic).
varlogp_nojac	PyTensor scalar of log-probability of the unobserved random variables (excluding deterministic) without jacobian term.

__init__(n_order=3, n_changepoints_trend=10, prior_sigma=5, trend_component=None, seasonality_component=None, sample_kwargs=None)

Parameters:

• sample_kwargs (Optional[Dict[str, Any]]) – A dictionary of kwargs that get unpacked and passed to the pymc.sample() function. Defaults to an empty dictionary.

• n_order (int)

• n_changepoints_trend (int)

• prior_sigma (float)

• trend_component (Any | None)

• seasonality_component (Any | None)

classmethod __new__(*args, **kwargs)