Dear Sir/Sirs,
To discriminate between two parametric (known) distribution, say f_0 and
f_1, we are interested to employ the Kullback-Leibler information. After
calculating K(f_0, f_1), how can we decide to choose f_0 or f_1? For
example, If K=4.5, what is your suggestion to reject (or accept) f_0, in 95%
level of confidence?
Any comments?
Thanks in advance,
ALI
|