A new extended Cardioid model: an application to wind data

TL;DR

This paper introduces the Exponentiated Cardioid distribution, an extended model for circular data that captures asymmetry and multimodality, with properties, estimation methods, and application to wind data.

Contribution

The paper proposes the Exponentiated Cardioid distribution, extending the Cardioid model to better fit asymmetric and multimodal circular data, with derived properties and estimation techniques.

Findings

EC distribution outperforms Cardioid and von Mises in wind data application.

Estimation methods show good bias and MSE performance in simulations.

Model captures asymmetry and multimodality effectively.

Abstract

The Cardioid distribution is a relevant model for circular data. However, this model is not suitable for scenarios were there is asymmetry or multimodality. In order to overcome this gap, an extended Cardioid model is proposed, which is called Exponentiated Cardioid (EC) distribution. Besides, some of its properties are derived, such as trigonometric moments, kurtosis and skewness. A discussion about the modality and and expressions for the quantiles through approximations of the studied model are also presented. To fit the EC model, two estimation methods are presented based on maximum likelihood and quantile least squares procedures. The performance of proposed estimators is evaluated in a Monte Carlo simulation study, adopting both bias and mean square error as comparison criteria. Finally, the proposed model is applied to a dataset in the wind direction context. Results indicate that the EC distribution may outperform Cardioid and the von Mises distributions.