binomial_adstock#

pymc_marketing.mmm.transformers.binomial_adstock(x, alpha=0.5, *, l_max=12, normalize=False, mode=ConvMode.After, dim)[source]#

Binomial adstock transformation.

Binomial adstock assumes that the effect of one unit of spend at time \(0 \le t \le l_{max} + 1\) is given by

\[f(t) = \left(1 - \frac{t}{L+ 1} \right)^{\left(\frac{1}{\alpha} - 1\right)}.\]

Notice that \(f(l_{max} + 1)=0\).

(Source code, png, hires.png, pdf)

../../_images/pymc_marketing-mmm-transformers-binomial_adstock-1.png
Parameters:
  • x (Union[Variable, Sequence[Variable], ArrayLike, XTensorVariable, DataArray]) – Input tensor.

  • alpha (Union[Variable, Sequence[Variable], ArrayLike, XTensorVariable, DataArray]) – Retention rate of ad effect. Must be between 0 and 1.

  • l_max (int) – Maximum duration of carryover effect.

  • normalize (bool) – Whether to normalize the weights.

  • dim (str) – The dimension of the x input along which to perform the convolution.

  • mode (ConvMode) –

    The convolution mode determines how the convolution is applied at the boundaries of the input signal, denoted as “x.” The default mode is ConvMode.After.

    • ConvMode.After: Applies the convolution with the “Adstock” effect, resulting in a trailing decay effect.

    • ConvMode.Before: Applies the convolution with the “Excitement” effect, creating a leading effect

      similar to the wow factor.

    • ConvMode.Overlap: Applies the convolution with both “Pull-Forward” and “Pull-Backward” effects,

      where the effect overlaps with both preceding and succeeding elements.

Return type:

tensor

Returns:

Transformed tensor.