The Chandra X-ray Observatory (CXO) is generating a tremendous amount of multi-dimensional X-ray data of exceptional quality. Currently, astronomers analyze these data one X-ray source at a time, via model-fitting techniques, to determine source physical conditions. More efficient methods of spectral and temporal classification would greatly benefit analysis of observations of rich fields of X-ray sources, such as stellar clusters. A combination of techniques from the fields of multivariate statistics and pattern recognition may provide new insight into, as well as an improvement in the speed and accuracy of, the classification of stellar X-ray spectra. We are adapting and applying such techniques, in the context of analysis of CXO and X-ray Multi-Mirror Mission (XMM-Newton) imaging spectroscopy of star formation regions, to group pre-main-sequence X-ray sources into clusters based on spectral attributes. An automated spectral classification technique for the Orion Nebula Cluster (ONC) population of greater than 1000 X-ray emitting young stars has been
developed. As an initial test of the algorithm, deep CXO images of the ONC were analyzed. Clustering results are being compared with
known optical, infrared, and radio properties of the young stellar
population of the ONC, to assess the algorithm's ability to identify
groups of sources that share common attributes.