bass

Bass diffusion model for product adoption forecasting.

The recommended entry point is BassModel, which wraps the model in a ModelBuilder interface with standard .fit(), .save(), and .load() methods.

The lower-level create_bass_model() and BassPriors are still available for users who need the raw pm.Model without the class wrapper.

Examples

Fit a single-product model from an array of adoption counts:

import numpy as np
from pymc_marketing.bass import BassModel

model = BassModel()
idata = model.fit(data=np.random.poisson(lam=100, size=50))

Generate synthetic data from the prior, then fit the model:

import xarray as xr
import pymc as pm

ds = xr.Dataset({"T": np.arange(50)})
model = BassModel()
model.build_model(data=ds)

with model.model:
    prior = pm.sample_prior_predictive(draws=50, random_seed=42)
    y_sim = prior.prior["y"].sel(draw=0, chain=0)

idata = model.fit(data=y_sim.values)

Modules

data	Data conversion utilities for the Bass diffusion model.
model	Bass diffusion model for product adoption forecasting.