Skip to contents

Objective Function with Metric from Piepho

Source: R/metrics.R
objective_function_piepho.Rd

Create an objective function including even distribution and neighbor balance introduced by Piepho 2018.

Usage

objective_function_piepho(
  design,
  swap,
  spatial_cols,
  current_score_obj = NULL,
  swapped_items = NULL,
  pair_mapping = NULL,
  row_column = "row",
  col_column = "col",
  by = NULL,
  grid_index = NULL,
  ...
)

Arguments

design

A data frame representing the spatial information of the design

swap

A column name of the items to be swapped

spatial_cols

Column name(s) of the spatial factors

current_score_obj

A named list containing the current score

swapped_items

The items that had just been swapped

pair_mapping

A named vector of pairs generated from create_pair_mapping

row_column

Name of column representing the row of the design (default: "row")

col_column

Name of column representing the column of the design (default: "col")

by, grid_index

Optional grouping of plots into separate grids; see calculate_adjacency_score(). Neighbour balance sums across grids, while evenness of distribution is scored per grid and the scores summed, reported both in total and per grid. A grid with no treatment replicated inside it contributes 0.

...

Extra parameters passed from speed

Value

A function which returns a named list of numeric values with one required name score representing the score of the design (lower is better) with a signature function(design_df, swap, spatial_cols, ...). An objective may optionally return a components element: a named numeric vector of the additive pieces that sum to score (e.g. c(adjacency = ..., balance = ...)). When present, summary.design() reports this as a faithful score decomposition. See signature details in objective_function_signature.

References

Piepho, H. P., Michel, V., & Williams, E. (2018). Neighbor balance and evenness of distribution of treatment replications in row-column designs. Biometrical journal. Biometrische Zeitschrift, 60(6), 1172-1189. https://doi.org/10.1002/bimj.201800013

See also

objective_function(), create_pair_mapping()

Examples

design_df <- initialise_design_df(
  items = c(1, 2, 2, 1, 3, 3, 1, 3, 3),
  nrows = 3,
  ncols = 3
)

pair_mapping <- create_pair_mapping(design_df$treatment)
objective_function_piepho(design_df, "treatment", c("row", "col"), pair_mapping = pair_mapping)
#> $score
#> [1] 16.36667
#> 
#> $ed
#> $ed$`1`
#> $ed$`1`$`2`
#> $ed$`1`$`2`$msts
#> $ed$`1`$`2`$msts$`2`
#> [1] 1
#> 
#> 
#> $ed$`1`$`2`$min_mst
#> [1] 1
#> 
#> $ed$`1`$`2`$min_items
#> [1] "2"
#> 
#> 
#> $ed$`1`$`4`
#> $ed$`1`$`4`$msts
#> $ed$`1`$`4`$msts$`3`
#> [1] 3
#> 
#> 
#> $ed$`1`$`4`$min_mst
#> [1] 3
#> 
#> $ed$`1`$`4`$min_items
#> [1] "3"
#> 
#> 
#> $ed$`1`$`3`
#> $ed$`1`$`3`$msts
#> $ed$`1`$`3`$msts$`1`
#> [1] 2
#> 
#> 
#> $ed$`1`$`3`$min_mst
#> [1] 2
#> 
#> $ed$`1`$`3`$min_items
#> [1] "1"
#> 
#> 
#> 
#> 
#> $ed_per_grid
#>         1 
#> 0.1666667 
#> 
#> $bal
#> [1] 8
#> 
#> $adj
#> [1] 7
#> 
#> $nb
#> $nb$nb
#> 1,1 1,2 1,3 2,2 2,3 3,3 
#>   2   1   2   1   2   4 
#> 
#> $nb$max_nb
#> [1] 4
#> 
#> $nb$max_pairs
#> [1] "3,3"
#> 
#> $nb$var
#> [1] 1.2
#> 
#> 
#> $components
#> neighbour_balance even_distribution           balance         adjacency 
#>         1.2000000         0.1666667         8.0000000         7.0000000 
#> 
# usage in speed, speed(..., obj_function = objective_function_piepho, pair_mapping = pair_mapping)