Tuning grids

trop_grid() gives the candidate values of the three tuning parameters. These parameters are the decay of the time weights, the decay of the unit weights, and the nuclear-norm penalty. A value of 0 for time or unit gives equal weights. A value of Inf for nn removes the low-rank term. Then the model has only two-way fixed effects.

Usage

trop_grid(
  time = c(0, 0.1, 0.5, 1, 2, 5),
  unit = c(0, 0.1, 0.5, 1, 2, 5),
  nn = c(0.01, 0.1, 1, 10, Inf)
)

Arguments

time

The candidate values for the decay of the time weights. The values must be finite and 0 or more.

unit

The candidate values for the decay of the unit weights. The values must be finite and 0 or more.

nn

The candidate values for the nuclear-norm penalty. The values must be 0 or more. You can use Inf.

Value

A list of class trop_grid with the numeric vectors time, unit, and nn.

Examples

trop_grid(nn = c(0.1, 1, Inf))
#> $time
#> [1] 0.0 0.1 0.5 1.0 2.0 5.0
#> 
#> $unit
#> [1] 0.0 0.1 0.5 1.0 2.0 5.0
#> 
#> $nn
#> [1] 0.1 1.0 Inf
#> 
#> attr(,"class")
#> [1] "trop_grid"