The ability to manufacture accurate parts in single point incremental forming is dependent on the capability to properly predict accuracy response surfaces of individual features and feature interaction combinations formed using uncompensated tool paths. Recent studies show that the accuracy profiles obtained are dependent on the choice of material used for forming, in terms of magnitude, geometric shape and nature of errors (under forming and over forming). In this paper, an attempt is made to capture the effect of material properties on the accuracy response surfaces. The response surfaces are modeled using Multivariate Adaptive Regression Splines (MARS), which is a non-parametric multivariate regression technique that helps generating continuous response surfaces. The MARS functions are based on process and feature specific geometric parameters. A set of features and feature interactions for which the response surface dependence on material properties is well predicted is used to illustrate the applicability of the MARS method for predicting the accuracy. An in-process stereo camera system is used to measure the displacement fields for different materials using digital image correlation (DIC) and understand the material dislocation mechanism. Improvements in accuracy for different sheet metal materials based on the predicted response surfaces are then discussed.