Advancing Machine Learning for Stellar Activity and Exoplanet Period Rotation - Insights from Kepler Light Curves

Aug 2, 2026

The study applies machine learning models to estimate stellar rotation periods from corrected Kepler light-curve data, addressing accuracy limits of traditional techniques under noise and intrinsic stellar variability. The workflow uses initial period estimates from LS-Periodogram and Transit Least Squares, then trains, validates, and tests models using multiple Kepler IDs. Decision Tree, Random Forest, k-Nearest Neighbors, and Gradient Boosting are compared, alongside a Voting Ensemble. The Voting Ensemble achieves the best RMSE, while Random Forest is similarly accurate; Gradient Boosting underperforms. Predicted rotation periods closely match photometric reference periods, indicating ensemble ML as an effective route to support stellar astrophysics and exoplanet research.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Introduction

Stellar rotation periods and exoplanets

Challenges from noise, artifacts, and variability

Machine learning methods for light-curve analysis

Astronomy & Astrophysics manuscript no. Fazel_2024 ©ESO 2024 September 10, 2024 [ astro-ph . SR] 9 Sep 2024 Advancing Machine Learning for Stellar Activity and Exoplanet Period Rotation Insights from Kepler Light Curves Fatemeh Fazel Hesar⋆1, 2, 3 , Bernard Foing2, 3,, Ana M. Heras4,, Mojtaba Raouf2, 3,, Victoria Foing3,, Shima Javanmardi 1,, and Fons J. Verbeek 1, 1 Leiden Institute of Advanced Computer Science (LIACS), Leiden University, Einsteinweg 55, 2333 CC Leiden, The Netherlands e-mail: [f.fazel.hesar@liacs.leidenuniv.nl](f.fazel.hesar@liacs.leidenuniv.nl) 2 Leiden Observatory, Leiden University, P.O. Box 9513, 2300 RA Leiden, Netherlands 3 LUNEX & Eurospacehub-Green academy &, ESA BIC, Noordwijk, Netherlands 4 Directorate of Science, European Space Research and Technology Center (ESA-ESTEC), Keplerlaan 1, NL-2201 AZ, Noordwijk, Netherlands Received Aug 29, 2024; accepted September, 2024 ABSTRACT This study applied machine learning models to estimate stellar rotation periods from corrected light curve data obtained by the NASA Kepler mission. Traditional methods often struggle to estimate rotation periods accurately due to noise and variability in the lightcurve data. The workflow involved using initial period estimates from the LS-Periodogram and Transit Least Squares techniques, followed by splitting the data into training, validation, and testing sets. We employed several machine learning algorithms, including Decision Tree, Random Forest, K-Nearest Neighbors, and Gradient Boosting, and also utilized a Voting Ensemble approach to improve prediction accuracy and robustness. The analysis included data from multiple Kepler IDs, providing detailed metrics on orbital periods and planet radii. Performance evaluation showed that the Voting Ensemble model yielded the most accurate results, with an RMSE approximately 50% lower than the Decision Tree model and 17% better than the K-Nearest Neighbors model. The Random Forest model performed comparably to the Voting Ensemble, indicating high accuracy. In contrast, the Gradient Boosting model exhibited a worse RMSE compared to the other approaches. Comparisons of the predicted rotation periods to the photometric reference periods showed close alignment, suggesting the machine learning models achieved high prediction accuracy. The results indicate that machine learning, particularly ensemble methods, can effectively solve the problem of accurately estimating stellar rotation periods, with significant implications for advancing the study of exoplanets and stellar astrophysics. arXiv :2409 .05482v1 Key words. Exoplanets – Machine Learning – Kepler Mission – Stellar Rotation Periods – Period Rotation – Light Curves 1. Introduction Accurately determining stellar rotation periods from photometric light curve data is essential in studying exoplanets and stellar astrophysics. Stellar rotation periods reveal important information about the physical properties and evolutionary paths of stars, which is important for characterizing exoplanetary systems (Aigrain et al. 2015; Reinhold & Gizon 2015) . Applied machine learning models, including Random Forest and Gradient Boosting, have been used to estimate rotation periods from Kepler light curves, demonstrating the potential of these methods. In addition high-resolution spectroscopy is an effective technique for estimating the rotation periods of late-type stars, as demonstrated in earlier studies (Char & Foing 1993) . The study of exoplanets, particularly the detection and characterization of their transits, has emerged as a cornerstone in modern astrophysics. Exoplanet transits, the periodic dimming of a star’s brightness as a planet passes in front of it, offer a unique opportunity to probe the physical properties, composition, and potential habitability of distant worlds beyond our solar system (Seager & Mallén-Ornelas 2003) . ⋆ [f.fazel.hesar@liacs.leidenuniv.nl](f.fazel.hesar@liacs.leidenuniv.nl)

FAQs

How does the study estimate stellar rotation periods from Kepler data?

It uses corrected Kepler light curves and trains machine learning models to predict rotation periods, starting from initial estimates obtained via LS-Periodogram and Transit Least Squares.

Which machine learning models are evaluated, and how is performance compared?

Decision Tree, Random Forest, k-Nearest Neighbors, and Gradient Boosting are tested, with a Voting Ensemble used to improve accuracy. Performance is evaluated using RMSE and comparison to photometric reference periods.

Why do ensemble methods perform better in this work?

The results show the Voting Ensemble yields the lowest RMSE and close alignment with reference periods, indicating improved prediction accuracy and robustness over single-model approaches.

Aug 2, 2026

Upload by Valentina