Message-ID: <CAPD4hGCg3DJ4M1+KKeNzQrWtLECEfEGFYFN0B2bSxroj8mTcag@mail.gmail.com>
Date: 2014-10-08T03:11:55Z
From: Gmail
Subject: SPA test in ttr Test
Hi, All,
I am using dataSnoop in the ttr Test package to do superior predict ability
test for my models. And i had the following errors:
spa <- dataSnoop(a,bSamples=3,test="SPA")
Preparing to Analyze 24 Parameterizations of TTR
Error in na.fail.default(as.ts(x)) : missing values in object
In addition: Warning messages:
1: In log(x) : NaNs produced
2: In log(x) : NaNs produced
where a is the difference between the base models loss function and the
loss function of other models. Should "a " be always positive?
Best,
Jun
[[alternative HTML version deleted]]