denis.missing {agridat} | R Documentation |
Grain yield was measured on 5 genotypes in 26 environments. Missing values were non-random, but structured.
env
environment, 26 levels
gen
genotype factor, 5 levels
yield
yield
Denis, J. B. and C P Baril, 1992, Sophisticated models with numerous missing values: The multiplicative interaction model as an example. Biul. Oceny Odmian, 24–25, 7–31.
Used with permission of Jean-Baptists Denis.
H P Piepho, (1999) Stability analysis using the SAS system, Agron Journal, 91, 154–160. http://doi.og/10.2134/agronj1999.00021962009100010024x
data(denis.missing) dat <- denis.missing # view missingness structure if(require(reshape2)){ acast(dat, env~gen, value.var='yield') } require(lattice) redblue <- colorRampPalette(c("firebrick", "lightgray", "#375997")) levelplot(yield ~ gen*env, data=dat, col.regions=redblue, main="denis.missing - incidence heatmap") # stability variance (Table 3 in Piepho) require("nlme") m1 <- lme(yield ~ -1 + gen, data=dat, random= ~ 1|env, weights = varIdent(form= ~ 1|gen), na.action=na.omit) svar <- m1$sigma^2 * c(1, coef(m1$modelStruct$varStruct, unc = FALSE))^2 round(svar, 2) ## G5 G3 G1 G2 ## 39.25 22.95 54.36 12.17 23.77 # m2 <- asreml(yield ~ gen, random = ~ env + at(gen):env, data=dat)