A new non parametric estimator for Pdf based on inverse gamma distribution. Issue 23 (1st December 2016)

• Record Type: Journal Article
• Authors: Mousa, A. M.; Hassan, M. Kh.; Fathi, A.
• Is Part Of: Communications in statistics. Volume 45:Issue 23(2016)
• Page Start: 7002
• Page End: 7010
• Publication Date: 2016-12-01
• Subjects: boundary bias -- inverse gamma kernel -- density estimation -- bandwidth.; 62G07, 62G08; Mathematical statistics -- Periodicals; Mathematics; Statistics; 519.2
• DOI: 10.1080/03610926.2014.972575
• Languages: English
• ISSNs: 0361-0926

Abstract

ABSTRACT: The non parametric approach is considered to estimate probability density function (Pdf) which is supported on(0, ∞). This approach is the inverse gamma kernel. We show that it has same properties as gamma, reciprocal inverse Gaussian, and inverse Gaussian kernels such that it is free of the boundary bias, non negative, and it achieves the optimal rate of convergence for the mean integrated squared error. Also some properties of the estimator were established such as bias and variance. Comparison of the bandwidth selection methods for inverse gamma kernel estimation of Pdf is done.