Local polynomial regression for circular predictors

Marco Di Marzio , Agnese Panzera , Charles C. Taylor

We consider local smoothing of datasets where the design space is the -dimensional torus and the response variable is real-valued. Our purpose is to extend least squares local polynomial fitting to this situation. We give both theoretical and empirical results.

Mots clés: 62G07; 62G08; 62G20

Marco Di Marzio, Agnese Panzera, Charles C. Taylor. Local polynomial regression for circular predictors. Statistics and Probability Letters, 2009, 79 (19), pp.2066. ⟨10.1016/j.spl.2009.06.014⟩. ⟨hal-00573471⟩