Robust Switching Regressions Using the Laplace Distribution

This paper presents a robust method for dealing with switching regression problems. Regression models switch-points are broadly employed in diverse areas. Many traditional methods regressions can falter the presence of outliers or heavy-tailed distributions because modeling assumptions Gaussian errors. The outlier corruption datasets is often unavoidable. When misapplied, assumption lead to incorrect inference making. Laplace distribution known as longer-tailed alternative normal and connected least absolute deviation criterion. We propose model distributed To advance robustness, we extend fuzzy class create algorithm named FCL through classification maximum likelihood procedure. robustness properties resistance against high-leverage discussed. Simulations sensitivity analyses illustrate effectiveness superiority proposed algorithm. experimental results indicate that much more than EM-based Furthermore, Laplace-based time-saving t-based Diverse real-world applications demonstrate practicality approach.