Scalar-on-Function Relative Error Regression for Weak Dependent Case

Analyzing the co-variability between Hilbert regressor and scalar output variable is crucial in functional statistics. In this contribution, kernel smoothing of Relative Error Regression (RE-regression) used to resolve problem. Precisely, we use relative square error establish an estimator Hilbertian regression. As asymptotic results, observations are assumed be quasi-associated, demonstrate almost complete consistency constructed estimator. The feasibility model as a predictor time series data discussed. Moreover, give some practical ideas for selecting parameter based on bootstrap procedure. Finally, empirical investigation performed examine behavior RE-regression estimation its superiority practice.