The aim of this project was to developed a pattern recognition system for hand-written digits. The system should be able to recognize and classify digits correctly based on the 3-dimensional (3-D) location of the time series. The 3-D location data have been obtained using a LeapMotion sensor and digits are written as free hand strokes in the air. This work presents singular value decomposition (SVD) based feature extraction technique and a classifier that performs a Bayesian classification with a Gaussian distribution model to identify handwritten 3-D digits.