Development of logistic regression models to classify seat fit 2020-01-0869

The digital evaluation process of vehicle-seat dimensions is an efficient and cost-effective way to achieve better comfort and proper fit. The present study is intended to quantify the statistical relationships between seat dimensions (e.g., insert width and bolster height defined at SAE J2732) and subjective seat fit (e.g., too tight, right fit, or too wide). Subjective fit evaluations for 45 different vehicle seats were collected, and the corresponding vehicle seat dimensions at various cross-sectional planes were collected. The best subset logistic regression analyses were applied to quantify the relationships between the collected subjective fit and seat dimensions at each cross-sectional plane. As a result, significant seat dimensions on the subjective fit were identified and their statistical relationships were quantified as regression coefficients. The developed logistic models showed 80% - 90% as an overall classification accuracy for learning datasets, and around 80% accuracy for testing datasets with five-fold cross-validations. The developed models would be particularly useful to identify optimal seat dimensions providing proper fit, reduce development cost for an automobile seat, and increase work efficiency in the digital evaluation process of an automobile seat.