If you convert your data to a groupedData object you can specify the ordering with respect to an outer factor, which could be the stratum East or West in your case.
Dear All, I've installed Splus 6 for Windows I discover that the Matrix library that (I think) was supported under earlier versions, is not supported in the latest Windows version. Is there a better
Dear all, I had something bizarre when I tried to get the inverse of the same matrix with S-plus6.2 and R. 0.004000337),nrow = 2, byrow=TRUE) With S-plus: Problem in solve.qr(a): apparently singular
Quick question. In the output for a GLMM object in the lme4 library in R, there is a line that reads: Estimated scale (compare to 1) 0.9370735 Can I think of this as extra-binomial dispersion? If so
As Spencer indicates in his reply the problem is that the variables xx0, xx1, yy0, etc. are not accessible during the evaluation of garch. If you evaluate the individual lines at the top level in the
The test performed by the anova function should be conservative in the sense that the p-value returned will be an upper bound on the true p-value you would obtain via simulation of the null model. Mo
If you want to have complete control you can avoid the selfStart function entirely and determine your own starting estimates for the parameters. Because the asymptotic regression model has two condit
Probably not. The computational methods in the lme function are optimized for strictly nested grouping factors. In an observational data set of this size it is unlikely that strict nesting will hold,
I did this in R, which is why the logical values print as TRUE and FALSE, but I believe the same will work in S-PLUS. The idea is to apply the cumprod (cumulative product) function across each row. [
conLin is an internal representation of a linear mixed-effects model used in the lme step. (The nlme function alternates between lme steps and penalized nonlinear least squares steps when performing
To fit the model you want you will need to call nlme with a formula for the model and supply your own starting estimates for the fixed-effects parameters. The advantage of using an nlsList object as
The rational function you are fitting is a partially linear model because the coefficients a0, a1, a2, and a3 occur linearly in the model. You can reduce the complexity of the optimization by using t
In exchanging some mail with Bert Gunter I realized that you can't fit your model because the estimates are not unique. Multiply all the a's by 2 and all the b's by 2 and you get the same fit. You ca
Because it is not always meaningful. The R^2 value for a linear model results from comparing the linear model fit to that of a model with only a constant term. For a nonlinear model we cannot easily