Poisson Point Process with Functional Covariates to Analyse Star-Clusters Data

We propose a three-dimensional Poisson point process model that accounts for functional covariates observed at event locations. Each functional covariate is represented through Functional Principal Component Analysis, providing a set of scores which can be used as covariates in the model specification. The effect of a functional covariate on the intensity of the process is interpretable both by reconstructing a smooth coefficient function from the estimated score effects and by considering it as a multiplicative effect on the baseline intensity. We show the methodology by an application on young star clusters data.