Higher dimensional kernel methods

Abstract

The multivariate kernel density estimator (MKDE) for the analysis of data in more than one dimension is presented. This removes the cumbersome nature associated with the interpretation of multivariate results when compared with most common multivariate schemes. The effect of varying the window width in MKDE with the attendant consequence of distortion in shape especially when the window width is large and when the kernel itself does not fit into the family to which the observations are drawn is also examined.

Journal of the Nigerian Association of Mathematical Physics Vol. 9 2005: pp. 351-356