Obtaining accurate body measurements is a critical step when designing products to fit the human body. Compared to traditional manual methods, 3D body scanning has fundamentally enhanced the accessibility of the body, however, the datasets extracted from 3D body scans often have missing values. Recently, the applications of data-driven machine learning (ML) methods in anthropometrics studies and clothing-related work have been increasing. However, there has been limited research on exploring if missing data and difficult-to-extract measurements from 3D scans could be predicted accurately and efficiently by using ML methods. Therefore, this exploratory study investigates the potential use of four mainstream ML methods in improving the usefulness of a 3D body scan dataset.