Portal de Periódicos da CAPES

A new asymmetric ϵ -insensitive pinball loss function based support vector quantile regression model

2020; Elsevier BV; Volume: 94; Linguagem: Inglês

10.1016/j.asoc.2020.106473

1872-9681

Pritam Anand, Reshma Rastogi, Suresh Chandra,

Financial Risk and Volatility Modeling

In this paper, we propose a novel asymmetric ϵ-insensitive pinball loss function for quantile estimation. There exists some pinball loss functions which attempt to incorporate the ϵ-insensitive zone approach in it but, they fail to extend the ϵ-insensitive approach for quantile estimation in true sense. The proposed asymmetric ϵ-insensitive pinball loss function can make an asymmetric ϵ- insensitive zone of fixed width around the data and divide it using τ value for the estimation of the τth quantile. The use of the proposed asymmetric ϵ-insensitive pinball loss function in Support Vector Quantile Regression (SVQR) model improves its prediction ability significantly. It also brings the sparsity back in SVQR model. Further, the numerical results obtained by several experiments carried on simulated and real world datasets empirically show the efficacy of the proposed 'ϵ-Support Vector Quantile Regression' (ϵ-SVQR) model over other existing SVQR models.