sensitivity

Sensitivity namespace — sensitivity analysis plots.

SensitivityPlots provides three methods to visualise the results of a sensitivity sweep run via SensitivityAnalysis:

• analysis() — raw effect curves from the input sweep

• uplift() — uplift relative to a baseline (with reference lines)

• marginal() — marginal effects along the sweep

The class is normally accessed through the mmm.plots.sensitivity shortcut on a fitted MMM instance, but it can also be constructed directly from any MMMIDataWrapper.

Examples

# sensitivity analysis
sweeps = np.linspace(0.1, 2.0, 100)
mmm.sensitivity.run_sweep(
    sweep_values=sweeps,
    var_input="channel_data",
    var_names="channel_contribution",
    extend_idata=True,
)

sp = mmm.plots.sensitivity
fig, axes = sp.analysis()

# uplift curve
ref = mmm.idata.posterior.channel_contribution.sum(["channel", "date"]).mean(
    ["chain", "draw"]
)
mmm.sensitivity.compute_uplift_curve_respect_to_base(
    results=mmm.idata.sensitivity_analysis["x"],
    ref=ref,
    extend_idata=True,
)

fig, axes = sp.uplift(aggregation={"sum": "channel"}, figsize=(10, 4))


# marginal contribution curve
mmm.sensitivity.compute_marginal_effects(
    results=mmm.idata.sensitivity_analysis["uplift_curve"],
    extend_idata=True,
)

fig, axes = sp.marginal(aggregation={"sum": "channel"})

Classes

SensitivityPlots(data)

Sensitivity analysis plots (effect, uplift, and marginal curves).