Random forest automated supervised classification of Hipparcos periodic variable stars

Abstract

We present an evaluation of the performance of an automated classification of the Hipparcos periodic variable stars into 26 types. The sub‐sample with the most reliable variability types available in the literature is used to train supervised algorithms to characterize the type dependencies on a number of attributes. The most useful attributes evaluated with the random forest methodology include, in decreasing order of importance, the period, the amplitude, the V−I colour index, the absolute magnitude, the residual around the folded light‐curve model, the magnitude distribution skewness and the amplitude of the second harmonic of the Fourier series model relative to that of the fundamental frequency. Random forests and a multi‐stage scheme involving Bayesian network and Gaussian mixture methods lead to statistically equivalent results. In standard 10‐fold cross‐validation (CV) experiments, the rate of correct classification is between 90 and 100 per cent, depending on the variability type. The main mis‐classification cases, up to a rate of about 10 per cent, arise due to confusion between SPB and ACV blue variables and between eclipsing binaries, ellipsoidal variables and other variability types. Our training set and the predicted types for the other Hipparcos periodic stars are available online.

The Catalina Surveys Southern periodic variable star catalogue

“…This set is similar to the best set of parameters found by Dubath et al (2011). However, there are some clear differences.…”

“…In contrast, RRab's generally have M t > 0.5 (and longer periods), while RRc's have M t values near 0.4. The effect of M t is similar to that of non-parametric skew, which is effective in separating eclipsing binaries from other types of variables (see for example Richards et al 2011;Dubath et al 2011). However, M t is also useful for separating other types of variables from each other.…”

“…This situation contrasts with many automated classification experiments, where in many cases the classification has been undertaken using a pre-selected catalogue of periodic variables (e.g. Dubath et al 2011;Richards et al 2012). In such cases, one can assume that almost all of the input sources will indeed be periodic variables.…”

“…With this in mind, over the past decade a number of variability surveys have investigated applying automated classification to such sources. For example, Dubath et al (2011) applied Random Forest classification to Hipparcos data, while Richards et al (2012) applied the same general technique to techniques to automatically classify variable stars from the ASAS survey. In contrast, Palaversa et al (2013) used support vector machine (Cortes & Vapnik 1995) based machine learning to classify periodic variables from the LINEAR survey, and Elorrieta et al (2016) have found that the AdaBoost technique (Freund & Schapire 1997) provides encouraging results in the case of the NIR light curves obtained by the Vista Variables in the Vía Láctea survey (Minniti et al 2010).…”

ASTErIsM: application of topometric clustering algorithms in automatic galaxy detection and classification

“…It has been successfully applied to different astrophysical subjects, such as classification of periodicity in variable stars (Dubath et al 2011). Trees are constructed from bootstrapped samples of the original training set.…”