Data mining techniques on astronomical spectra data – II. Classification analysis

ABSTRACT Classification is valuable and necessary in spectral analysis, especially for data-driven mining. Along with the rapid development of surveys, a variety classification techniques have been successfully applied to astronomical data processing. However, it difficult select an appropriate method practical scenarios due different algorithmic ideas characteristics. Here, we present second work mining series – review techniques. This also consists three parts: systematic overview current literature, experimental analyses commonly used algorithms, source codes this paper. First, carefully investigate methods literature organize these into ten types based on their ideas. For each type algorithm, analysis organized from following perspectives. (1) applications usage frequencies are summarized; (2) basic introduced preliminarily analysed; (3) advantages caveats algorithm discussed. Secondly, performance algorithms unified sets analysed. Experimental selected LAMOST survey SDSS survey. Six groups designed characteristics, qualities, volumes examine algorithms. Then scores nine shown discussed analysis. Finally, written python manuals improvement provided.