As far as my inexpert knowledge goes, all the extensive literature on kernel
density estimation deals with symmetrical kernels.
Apart from the possible theoretical interest in relaxing an apparently
arbitrary condition, I have an empirical problem for which I can think of
substantive arguments in favour of an asymmetric kernel.
Has this notion in fact been studied?
Julian Wells
OU Business School
The Open University
Walton Hall
Milton Keynes
MK7 6AA
United Kingdom
+44 1908 654658
|