Autoregressive Planet Search: Methodology & Application to The Kepler Mission

Open Access
Author:
Caceres, Gabriel
Graduate Program:
Astronomy and Astrophysics
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
October 04, 2018
Committee Members:
  • Eric D Feigelson, Dissertation Advisor
  • Eric D Feigelson, Committee Chair
  • Eric B Ford, Committee Member
  • Jason Thomas Wright, Committee Member
  • Jogesh Babu Gutti, Outside Member
  • Robin Bruce Ciardullo, Committee Member
Keywords:
  • Astronomy
  • Exoplanets
  • Astrostatistics
Abstract:
The detection of periodic signals from transiting exoplanets is often impeded by extraneous aperiodic photometric variability, either intrinsic to the star or arising from the measurement process. Frequently, these variations are autocorrelated, wherein later flux values are correlated with previous ones. In this work we present the methodology of the Autoregessive Planet Search (ARPS) project which uses Autoregressive Integrated Moving Average (ARIMA) and related models that treat a wide variety of stochastic processes, as well as nonstationarity, to improve detection of new planetary transits. Providing a time series is evenly spaced or can be placed on an evenly spaced grid with missing values, these parametric models can prove very effective. We also introduce a planet-search algorithm to detect periodic transits in the residuals after the application of ARIMA models. Our matched-filter algorithm, the Transit Comb Filter (TCF), is closely related to the traditional Box-fitting Least Squares and provides an analogous periodogram. We show results from applying this methodology to 156,717 lightcurves from the \Kepler\ Prime Mission. A reduction in variability and autocorrelated noise is observed for the majority of the lightcurves examined. Scalar features from different stages of the analysis are then selected to use for machine learning classification. We use Random Forests for this task, in conjunction with Receiver Operating Characteristic (ROC) curves, to define discovery criteria for new, high fidelity candidates. We validate our method using Kepler Objects of Interest and report over a hundred potential new candidate transits.