A pattern-recognition algorithm has been used to search for edge operational space diagrams which discriminate well between the low-mode (L-mode) and high-mode (H-mode) states in the DIII-D [J. L. Luxon, R. Anderson, F. Batty et al., in Proceedings of the 11th Conference on Plasma Physics and Controlled Nuclear Fusion Research, 1986 (International Atomic Energy Agency, Vienna, 1987), Vol. I, p. 159] tokamak. The study uses a large database of electron temperatures Te, electron densities ne, electron pressures Pe, and their gradients, obtained for a fixed magnetic configuration. The data are evaluated at the edge of the plasma in the region where the H-mode transport barrier forms. An inductive classification algorithm shows that operational space diagrams which include either ∇Pe or ∇Te provide significantly better discrimination between the L-mode and H-mode states than diagrams based on Te and ne alone. These conclusions are based on standard metrics of pattern recognition. The successful application of a classification algorithm to this simple system provides confidence that classification techniques can be applied to larger and more complex databases of edge parameters. Moreover, these results suggest that edge pressure or temperature gradients should be considered on an equal footing with edge temperature as potential control parameters for the H-mode transition.