Additive Model

Additive models are a class of interpretable machine learning models that represent the output as a sum of individual functions, each depending on a single input feature. Current research focuses on enhancing their accuracy and scalability through various architectures, including neural additive models (NAMs), tensor polynomial additive models (TPAMs), and variants employing piecewise linear approximations or hierarchical structures to handle high-dimensional or complex data. This focus on interpretability, coupled with efforts to improve predictive performance, makes additive models valuable for applications requiring both accuracy and transparent decision-making, particularly in domains like finance, healthcare, and energy forecasting.