DiagnosticsPlots#

class pymc_marketing.mmm.plotting.diagnostics.DiagnosticsPlots(data)[source]#

Time-series diagnostic plots for fitted MMM models.

Provides six methods to visualize model fit and residuals:

  • posterior_predictive — Posterior predictive time series with HDI.

  • prior_predictive — Prior predictive time series with HDI.

  • residuals — Residuals (target − predictions) over time.

  • residuals_distribution — Posterior distribution of residuals.

  • posterior — 1-D marginal KDE distributions of posterior variables.

  • prior_vs_posterior — Overlaid prior and posterior KDE distributions.

Parameters:
dataMMMIDataWrapper

Validated wrapper around the fitted model’s InferenceData.

Methods

DiagnosticsPlots.__init__(data)

DiagnosticsPlots.posterior([var_names, ...])

Plot 1-D marginal KDE distributions for one or more posterior variables.

DiagnosticsPlots.posterior_predictive([...])

Plot time series from the posterior predictive distribution.

DiagnosticsPlots.prior_predictive([...])

Plot time series from the prior predictive distribution.

DiagnosticsPlots.prior_vs_posterior([...])

Overlay prior and posterior 1-D marginal KDE distributions.

DiagnosticsPlots.residuals_distribution([...])

Plot the posterior distribution of residuals using arviz-plots.

DiagnosticsPlots.residuals_over_time([...])

Plot residuals (target − posterior predictions) over time.