A two-factor approach to fit technically makes mass customization possible by exchanging body-shape data sets. For simple styles, this is indeed the case. For complex styles, however, design-for-body-shape pattern techniques present challenges.

CAD parameterization is effective at linear grading. Complex styles may use dart manipulation or shaping strategies specific to unique morphology. The body-shape specific pattern technique changes pattern geometry in unpredictable ways. Machine learning has much to offer here since the practice is grounded in logic.

To provide sufficient machine training data the design-for-body-shape technique can be developed on body-shape blocks to optimize machine training data. This reduces an unrealistic global number of human shapes to a manageable size from which design teams can begin to build logic-based strategies. In contrast, heuristic machine training data will inevitably produce unpredictable results.