MMMIDataWrapper#

class pymc_marketing.data.idata.mmm_wrapper.MMMIDataWrapper(idata, schema=None, validate_on_init=True)[source]#

Codified wrapper around InferenceData for MMM models.

Provides validated access to data and common transformations.

Parameters:

idataaz.InferenceData: InferenceData object from fitted MMM model
schemaMMMIdataSchema, optional: Schema to validate against. If None, validation skipped.
validate_on_initbool, default True: Whether to validate idata structure on initialization

Examples

>>> wrapper = MMMIDataWrapper(mmm.idata)
>>>
>>> # Access observed data
>>> observed = wrapper.get_target()
>>>
>>> # Get contributions in original scale
>>> contributions = wrapper.get_contributions(original_scale=True)

Methods

`MMMIDataWrapper.__init__`(idata[, schema, ...])
`MMMIDataWrapper.aggregate_dims`(dim, values, ...)	Aggregate multiple dimension values into one.
`MMMIDataWrapper.aggregate_time`(period[, method])	Aggregate data over time periods.
`MMMIDataWrapper.compute_posterior_summary`(var)	Compute summary statistics for a variable.
`MMMIDataWrapper.filter_dates`([start_date, ...])	Filter to date range, returning new wrapper.
`MMMIDataWrapper.filter_dims`(**dim_filters)	Filter by custom dimensions, returning new wrapper.
`MMMIDataWrapper.get_channel_contributions`([...])	Get channel contribution posterior samples.
`MMMIDataWrapper.get_channel_scale`()	Get channel scaling factor used during model fitting.
`MMMIDataWrapper.get_channel_spend`()	Get channel spend data with consistent access pattern.
`MMMIDataWrapper.get_contributions`([...])	Get all contribution variables in a single dataset.
`MMMIDataWrapper.get_roas`([original_scale])	Compute ROAS (Return on Ad Spend) for each channel.
`MMMIDataWrapper.get_target`([original_scale])	Get observed target data with consistent access pattern.
`MMMIDataWrapper.get_target_scale`()	Get target scaling factor used during model fitting.
`MMMIDataWrapper.to_original_scale`(var)	Transform variable from scaled to original scale.
`MMMIDataWrapper.to_scaled`(var)	Transform variable from original to scaled space.
`MMMIDataWrapper.validate`()	Validate idata structure against schema.
`MMMIDataWrapper.validate_or_raise`()	Validate idata structure, raising detailed exception if invalid.

Attributes

`channels`	Get channel coordinate.
`custom_dims`	Get all custom dimension names.
`dates`	Get date coordinate.