MultivarLinearReg

class causalpy.pymc_models.MultivarLinearReg

PyMC linear regression for multivariate outcomes (3D y) with covariance.

For use with InterruptedTimeSeries when multiple outcomes are specified (list of formulas). Accepts y with dims [“obs_ind”, “treated_units”, “outcomes”] and X with dims [“obs_ind”, “coeffs”]. Each (unit, outcome) has its own coefficient vector beta. Residuals across outcomes share a covariance matrix (LKJ prior), so you can estimate and interpret correlations between outcomes.

\[\begin{split}\beta &\sim \mathrm{Normal}(0, 50) \\ \Sigma &\sim \mathrm{LKJCholeskyCov}(\eta, n\_outcomes) \\ \mu_{i,u,o} &= X_i \cdot \beta_{u,o} \\ y_{i,u,\cdot} &\sim \mathrm{MvNormal}(\mu_{i,u,\cdot}, \Sigma)\end{split}\]

Methods

MultivarLinearReg.__init__([sample_kwargs, ...])
MultivarLinearReg.add_coord(name[, values, ...])	Register a dimension coordinate with the model.
MultivarLinearReg.add_coords(coords, *[, ...])	Vectorized version of Model.add_coord.
MultivarLinearReg.add_named_variable(var[, dims])	Add a random graph variable to the named variables of the model.
MultivarLinearReg.build_model(X, y, coords)	Define the PyMC model: regression + LKJ covariance between outcomes.
MultivarLinearReg.calculate_cumulative_impact(impact)
MultivarLinearReg.calculate_impact(y_true, ...)	Calculate the causal impact as the difference between observed and predicted values.
MultivarLinearReg.check_start_vals(start, ...)	Check that the logp is defined and finite at the starting point.
MultivarLinearReg.compile_d2logp([vars, ...])	Compiled log probability density hessian function.
MultivarLinearReg.compile_dlogp([vars, jacobian])	Compiled log probability density gradient function.
MultivarLinearReg.compile_fn(outs, *[, ...])	Compiles a PyTensor function.
MultivarLinearReg.compile_logp([vars, ...])	Compiled log probability density function.
MultivarLinearReg.copy()	Clone the model.
MultivarLinearReg.create_value_var(rv_var, ...)	Create a TensorVariable that will be used as the random variable's "value" in log-likelihood graphs.
MultivarLinearReg.d2logp([vars, jacobian, ...])	Hessian of the models log-probability w.r.t.
MultivarLinearReg.debug([point, fn, verbose])	Debug model function at point.
MultivarLinearReg.dlogp([vars, jacobian])	Gradient of the models log-probability w.r.t.
MultivarLinearReg.eval_rv_shapes()	Evaluate shapes of untransformed AND transformed free variables.
MultivarLinearReg.fit(X, y[, coords])	Draw samples from posterior, prior predictive, and posterior predictive distributions.
MultivarLinearReg.get_context([...])
MultivarLinearReg.initial_point([random_seed])	Compute the initial point of the model.
MultivarLinearReg.logp([vars, jacobian, sum])	Elemwise log-probability of the model.
MultivarLinearReg.logp_dlogp_function([...])	Compile a PyTensor function that computes logp and gradient.
MultivarLinearReg.make_obs_var(rv_var, data, ...)	Create a TensorVariable for an observed random variable.
MultivarLinearReg.name_for(name)	Check if name has prefix and adds if needed.
MultivarLinearReg.name_of(name)	Check if name has prefix and deletes if needed.
MultivarLinearReg.point_logps([point, ...])	Compute the log probability of point for all random variables in the model.
MultivarLinearReg.predict(X[, coords, ...])	Predict data given input data X.
MultivarLinearReg.print_coefficients(labels)	Print regression coefficients per outcome; residual covariance is in posterior 'cov'.
MultivarLinearReg.priors_from_data(X, y)	Generate priors dynamically based on the input data.
MultivarLinearReg.profile(outs, *[, n, ...])	Compile and profile a PyTensor function which returns outs and takes values of model vars as a dict as an argument.
MultivarLinearReg.register_data_var(data[, dims])	Register a data variable with the model.
MultivarLinearReg.register_rv(rv_var, name, *)	Register an (un)observed random variable with the model.
MultivarLinearReg.replace_rvs_by_values(...)	Clone and replace random variables in graphs with their value variables.
MultivarLinearReg.score(X, y[, coords])	R² per (treated_unit, outcome) for multivariate y.
MultivarLinearReg.set_data(name, values[, ...])	Change the values of a data variable in the model.
MultivarLinearReg.set_dim(name, new_length)	Update a mutable dimension.
MultivarLinearReg.set_initval(rv_var, initval)	Set an initial value (strategy) for a random variable.
MultivarLinearReg.shape_from_dims(dims)
MultivarLinearReg.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.
default_priors
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
supports_multivariate
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__(sample_kwargs=None, priors=None)

Parameters:

• sample_kwargs (dict, optional) – Dictionary of kwargs that get unpacked and passed to the pymc.sample() function. Defaults to an empty dictionary if None.

• priors (dict, optional) – Dictionary of priors for the model. Defaults to None, in which case default priors are used.

Return type:

None

classmethod __new__(*args, **kwargs)