Outlier Detection in Non-Linear Regression Analysis Based on the Normalizing Transformations

Abstract

The statistical technique for detecting outliers in non-linear regression analysis of non-Gaussian data based on the normalizing transformations and prediction intervals is proposed. The application of the technique is considered for detecting outliers in four-variate non-Gaussian data, which used for constructing the three-factor non-linear regression model based on the normalizing transformations, both multivariate and univariate. We demonstrate that the width of the non-linear regression prediction interval based on the Johnson four-variate transformation is less than after using the Johnson and decimal logarithm univariate transformations.