Mixing of air masses between different compartments of the atmosphere is one of the processes ruling atmospheric composition. The mixing process is commonly studied by using tracer-tracer correlations. Here, we generalize this approach by statistical classification methods based on a larger number of tracers to quantify mixing. From the 3-D resolution of our trace gas observations we are able to spatially resolve the observed mixing processes. This paper presents a matching flight-experiment of a filament of Asian monsoon air in the Upper Troposphere/ Lower Stratosphere (UTLS) off the North-American west coast by two flights of the High Altitude and Long Range research aircraft (HALO) conducted during the "Probing High Latitude Export of air from the Asian Summer Monsoon (PHILEAS)" campaign. In both flights the 3-D structure of the filament was revealed by tomographic observations by the Gimballed Limb Observer for Radiance Imaging of the Atmosphere (GLORIA) of five trace species (H₂O, PAN, CFC – 12, O₃, HNO₃). The observed tracer mixing ratios show evidence for a tropopause folding in connection to a Rossby wave breaking event. We show that the strongly perturbed atmospheric situation can not be decisively described by simple tracer-tracer correlations. By using a Bayesian Gaussian mixture model to cluster our observations by similarity we identify five classes of air masses: tropospheric air (both continental and maritime), Asian Summer Monsoon outflow (ASMO), mixed air and stratospheric air. Trajectory calculations are carried out to identify air masses which are observed in both flights. A measure of the mixing strengths of the mixing between both flights follows naturally from this classification. The unique 3-D observations allow us to reveal the spatial structure of the mixing processes in high detail. In particular, the mixing of ASMO air directly with stratospheric air and into the UTLS are shown. Comparing the classification to simulated artificial surface-origin tracers in the Chemical Lagrangian Model of the Stratosphere (CLaMS), we find strong evidence for distinctly correlated air masses to originate within different source regions within the Asian monsoon region.