1112.1290 (Noble P. Abraham)
Noble P. Abraham
We present a non-parametric statistical approach on the classification of
transient events based on photometric observations. The time series data of
blazars, CVs, Mira variables and supernovae from Catalina Real-Time Transient
Survey (CRTS) are used in the analysis discussed in this paper. For each class
of object, templates are generated with which a correlation analysis is carried
out and statistics derived. Based on the derived statistics the test object is
assigned a class. Templates were constructed using random sampling and
bootstrap sampling techniques. We discuss the trends of misclassification, in
the classification based on the templates constructed from the available
dataset, and on the ability of the templates to distinguish between the
classes.
View original:
http://arxiv.org/abs/1112.1290
No comments:
Post a Comment