Data Mining In Massive Spectral Data

with the astronomical spectra data grows rapidly, it becomes impossible for astronomers to read all the data manually, especially for some sky survey telescopes like SLOAN which will yield immense amounts of data every observational night. Automated astronomical data analysis software and system will be very necessary and useful. In this paper, a data mining application based on PCA (Principal Component Analysis) and LOF (Local Outlier Factor) is explored. Massive spectral data are clustered after dimension reduction by PCA and singular spectra candidates can be found out automatically. Some rare celestial body candidates are found out in massive spectra data that proves out method is feasible.