The XMM-Newton observatory is collecting a tremendous amount of X-ray imaging spectroscopy data. To deal with this huge volume of data, we are investigating more efficient methods to classify astronomical sources based purely on their X-ray spectra, and to understand the fundamental physical mechanisms responsible for X-ray emission. Multivariate statistics and pattern classification techniques are powerful tools to provide insight into the spectral similarities between a given target and its neighbors in the same observation. With this goal, we are developing approaches to classification of X-ray CCD spectra obtained by the XMM EPIC CCD instruments. Although X-ray CCD spectra have low resolution, they can be obtained in batches, whereas a high resolution spectrum can be only generated by the XMM RGS spectrometer for the brightest sources. Furthermore, X-ray CCD spectra can yield the relationship, if any, between the target source and other sources in the same field. The initial results are demonstrated by using a field centered on V1647 Ori, a young star that has recently displayed an accretion-driven optical, infrared and X-ray outburst. We applied Principle Component Analysis (PCA) to reduce the data dimensionality and Independent Component Analysis (ICA) to separate the CCD spectra as independently as possible. Then the Hierarchical Clustering classification method was employed to discriminate between this eruptive young star and other pre-main sequence X-ray sources in the field.