On-farm tricot evaluation of potatoes

On-farm evaluation of 3 potato genotypes per trial in Rwanda

Usage

data("desousa.potato")

Format

A data frame with 168 observations on the following 11 variables.

id

trial ID

project_name

project

gen_A

genotype 1 in each trial

gen_B

genotype 2

gen_C

genotype 3

start

start date

end

end date

yield_pos

highest-yielding genotype

yield_neg

lowest-yielding genotype

lon

longitude

lat

latitude

Details

The data come from "tricot" (triadic comparisons of technology) trials. On-farm potato evaluation trials conducted in Rwanda, beginning Sep/Oct 2020.

There are 11 genotypes evaluated in 168 field trials in an incomplete block design with 3 genotypes at each trial. Within each trial the highest-yield and lowest-yield genotypes are recorded.

Electronic data retrieved 2025.12.08 from https://zenodo.org/records/6286006. File: data/potato.csv Open license. Also at https://github.com/AgrDataSci/tricot-framework-rtb/tree/v0.1

Data curated by K.Wright

Source

de Sousa, K., van Etten, J., Manners, R. et al. (2024). The tricot approach: an agile framework for decentralized on-farm testing supported by citizen science. A retrospective. Agronomy for Sustainable Development, 44, 8. https://doi.org/10.1007/s13593-023-00937-1

References

None

Examples

if (FALSE) { # \dontrun{
  library(agridat)
  data(desousa.potato)
  dat <- agridat::desousa.potato

  # Convert each trial result to vector of ranks
  libs(gosset)
  datrnk <- gosset::rank_tricot(dat,
                                # Specify the columns containing the three genotypes
                                items=c("gen_A", "gen_B","gen_C"),
                                # Specify the columns containing the rankings
                                input=c("yield_pos", "yield_neg"),
                                group = TRUE)
  head(datrnk)

  # Basic PlacketLuce model using only the ranks
  libs(PlackettLuce)
  mod0 <- PlackettLuce(datrnk)
  
  # worth parameters (sum to 1) - probability of each item being ranked first
  coef(mod0, log = FALSE)
  plot(mod0, log=FALSE)

  # Compare pairs of entries, quasi standard errors
  # vertical line is at 1/11 (number of varieties)
  mod0qv <- qvcalc::qvcalc(mod0)
  # This is log scale.
  plot(mod0qv, ylab = "Worth (log)", main = "desousa.potato", las=2)

  # Could add weather covariates as in
  # https://www.sciencedirect.com/science/article/pii/S2352711023000985?via
} # }