Geometric goodness-of-fit measures have demonstrated strong potential to uncover structural patterns in classical data where each observation is a single point, by quantifying spatial relationships within clouds of points. Building on this foundation, Solís and Hernández (2023) introduced an index based on the area difference between the alpha shape of a two-dimensional projection of the data and its minimal bounding rectangle. In this study, we propose an extension of this geometric index to regression models involving symbolic data represented by intervals. Our approach analyzes rectangular clusters defined by the interval values of predictor and response variables, adapting the alpha shape methodology to address both the central tendency and variability inherent in interval observations. We introduce an interval geometric fit function that integrates alpha shapes built from the centers and ranges (In this paper, we use the term range to refer to the half-range of an interval, i.e., half the width of the interval.) of interval-valued data. This function, defined over the scale parameter \(\alpha \) , quantifies geometric goodness-of-fit by revealing both concordance and divergence between the location and variability components. The method is implemented in R and applied to benchmark symbolic datasets, demonstrating its ability to reveal geometric structures and provide insights that complement classical symbolic regression metrics such as the coefficient of determination. All developed functions are publicly available and designed for general use in symbolic data analysis.