Orthogonal least trimmed absolute deviation estimator for multiple linear errors-in-variables model

Orthogonal least trimmed absolute deviation (OLTAD) estimator of the multiple linear errors-in-variables (EIV) model is presented. We show that the OLTAD estimator has the high breakdown point and appropriate properties. A new decimal-integer-coded genetic algorithm(DICGA) and Fast-OLTAD method for solving OLTAD estimators are also proposed. Computational experiments of the OLTAD estimator of the multiple linear EIV model on benchmark data and synthetic data are provided. The results indicate that the DICGA and Fast-OLTAD methods perform well in dealing with high leverage outliers in reasonable computational time.