MARS (Multivariate Adaptive Regression Splines) is a companion to CART that focuses on the development and deployment of accurate and easy-to-understand regression models. The MARS model is designed to predict continuous numeric outcomes such the average monthly bill of a mobile phone customer or the amount that a shopper is expected to spend in a web site visit. MARS is also capable of producing high quality probability models for a yes/no outcome. A dramatic improvement over conventional stepwise and other automated regression tools, MARS performs variable selection, variable transformation, interaction detection, and self-testing, all automatically and at high speed. The MARS model is a regression but with automatically generated non-linearities and interactions included. A number of independent scientific studies have reported that MARS often outperforms neural networks in predictive accuracy while training from 100 to 1000 times faster.

MARS excels at finding thresholds and breaks in the relationships between a set of inputs and is thus ideal for detecting changes in the behavior of individuals or processes over time. Of all the Salford tools, MARS is the most adept at working with the small data sets frequently encountered in engineering contexts. MARS has also been involved in winning data mining competitions focused on large database customer relationship management (CRM) topics. Areas where MARS has exhibited very high-performance results include forecasting electricity demand for power generating companies, relating customer satisfaction scores to the engineering specifications of products, and presence/absence modeling in geographical information systems (GIS).