This contribution presents neuralGAM, an R package to fit Generalized Additive Models (GAMs) using neural networks as smooth function estimators, combining the flexibility of deep learning with the transparency of additive models.
The package supports several distributions and link functions, covering both regression and classification tasks, and provides built-in tools for model fitting, prediction, uncertainty estimation and graphical inspection of the estimated partial effects.
Presented on 4 September 2026 in the session GT SW I: Paquetes de R at the XLII Congreso Nacional de Estadística, Investigación Operativa y Ciencia de Datos (SEIO 2026), Santiago de Compostela.