agridat {agridat}
Description
This package contains datasets from published papers and books relating to agriculture, especially to field experiments.
Details
| name | gen | loc | reps | years | trt | other | model |
| aastveit.barley | 15 | 9 | Yr*Gen~Yr*Trait | pls | |||
| allcroft.lodging | 32 | 7 | percent | tobit | |||
| australia.soybean | 58 | 4 | 2 | 4-way, 6 traits | biplot | ||
| batchelor.apple | xy, uni | ||||||
| batchelor.lemon | xy, uni | ||||||
| batchelor.navel1 | xy, uni | ||||||
| batchelor.navel2 | xy, uni | ||||||
| batchelor.valencia | xy, uni | ||||||
| batchelor.walnut | xy, uni | ||||||
| besag.elbatan | 50 | 3 | xy | lm, gam | |||
| besag.met | 64 | 6 | 3 | xy, incblock | asreml, lme | ||
| bridges.cucumber | 4 | 2 | 4 | xy, latin, hetero | asreml | ||
| cochran.bib | 13 | 13 | BIB | aov, lme | |||
| corsten.interaction | 20 | 7 | |||||
| crowder.germination | 2 | 21 | 2 | glm | |||
| cox.stripsplit | 4 | 3,4,2 | aov | ||||
| denis.missing | 5 | 26 | lme | ||||
| diggle.cow | 4 | ts | |||||
| durban.competition | 36 | 3 | xy, competition | lm | |||
| durban.rowcol | 272 | 2 | xy | lm, gam, asreml | |||
| durban.splitplot | 70 | 4 | 2 | xy | lm, gam, asreml | ||
| eden.potato | 4 | 3 | 4-12 | xy, rcb, latin | aov | ||
| federer.tobacco | 8 | 7 | xy | lm | |||
| gathmann.bt | 2 | 8 | TOST | ||||
| gauch.soy | 7 | 7 | 4 | 12 | AMMI | ||
| gilmour.serpentine | 108 | 3 | xy, serpentine | asreml | |||
| gomez.fractionalfactorial | 2 | 6 | xy | lm | |||
| gomez.groupsplit | 45 | 3 | 2 | xy, 3 gen groups | aov | ||
| gomez.multilocsplitplot | 2 | 3 | 3 | nitro | aov, lmer | ||
| gomez.splitsplit | 3 | 3 | xy, nitro, mgmt | aov, lmer | |||
| gomez.stripplot | 6 | 3 | xy, nitro | aov | |||
| gomez.stripsplitplot | 6 | 3 | xy, nitro | aov | |||
| gomez.uniformity | xy, uni | aov | |||||
| graybill.heteroskedastic | 4 | 13 | hetero | ||||
| hanks.sprinkler | 3 | 3 | xy | asreml | |||
| hildebrand.systems | 14 | 4 | asreml | ||||
| hughes.grapes | 3 | 6 | binomial | lmer, aod, glmm | |||
| kempton.competition | 36 | 3 | xy, competition | lme AR1 | |||
| kempton.rowcol | 35 | 2 | uni, row-col | lmer | |||
| kempton.uniformity | uni | ||||||
| mcconway.turnip | 2 | 4 | 2,4 | hetero | aov, lme | ||
| mead.strawberry | 8 | 4 | |||||
| mercer.wheat | xy, uni | spplot | |||||
| pearl.kernels | chisq | ||||||
| rothamsted.brussels | 4 | 6 | |||||
| shafii.rapeseed | 6 | 14 | 3 | 3 | biplot | ||
| smith.uniformity3 | 4 | 3 | xy, uni | ||||
| stroup.nin | 56 | 4 | xy | asreml | |||
| stroup.splitplot | 4 | asreml, MCMCglmm | |||||
| student.barley | 2 | 51 | 6 | lmer | |||
| talbot.potato | 9 | 12 | 6 | Gen*Env~Gen*Trait | pls | ||
| theobald.covariate | 10 | 7 | 5 | cov | jags | ||
| thompson.cornsoy | 5 | 33 | corn/soy, repeated measures | aov | |||
| vargas.wheat1 | 7 | 6 | Gen*Yr~Gen*Trait, Yr*Gen~Yr*Cov | pls | |||
| vargas.wheat2 | 8 | 7 | Env*Gen~Env*Cov | pls | |||
| verbyla.lupin | 9 | 8 | 2 | xy, density | |||
| wedderburn.barley | 10 | 9 | percent | glm | |||
| williams.barley | uni | ||||||
| williams.cotton | uni | ||||||
| williams.trees | 37 | 6 | 2 | ||||
| wiebe.wheat | xy, uni | medianpolish, loess | |||||
| yan.winterwheat | 18 | 9 | biplot | ||||
| yates.missing | 10 | 3^2 | factorial | lm, pca | |||
| yates.oats | 3 | 6 | xy, nitro | lmer |
Abbreviations in the 'other' column include: xy = coordinates, uni = uniformity trial, pls = partial least squares, row-col = row-column design, ts = time series.
The original sources for these data use several different words to refer to genetics including line, cultivar, hybrid, variety, type, stock, and genotype. For simplicity and consistency, these datasets all use gen (genotype).
Box (1957) said, "I had hoped that we had seen the end of the obscene tribal habit practiced by statisticians of continually exhuming and massaging dead data sets after their purpose in life has long since been forgotten and there was no possibility of doing anything useful as a result of this treatment."
Clearly, massaging these 'dead' data sets will not lead to any of the genetics being released for commercial use. The value of these data is, however, multifold: 1. Validating published analyses (reproducible research). 2. Providing data for testing new analysis methods. 3. Illustrating the use of R. 4. Learning from history so as not to repeat it.
Some of the examples use the asreml package since it is the only option for fitting mixed models with complex variance structures to large datasets, and also the only option (even for small datasets) for modelling AR1xAR1 structures. The Discovery version of ASREML is free for people in academia (excluding commercial use) and for people in developing nations. This applies to both the stand-alone ASREML and the R package ASREML-R. Learn more at http://www.vsni.co.uk/software/asreml-discovery/. Commercial use requires a license: http://www.vsni.co.uk/downloads/asreml/.
References
Box G. E. P. (1957), Integration of Techniques in Process Development, Transactions of the American Society for Quality Control.
Documentation reproduced from package agridat, version 1.3. License: GPL-2