Best constant predictor under squared error

Two probe constants either side of the mean are evaluated; if either beats the mean the routine stops rather than returning a wrong optimum.

Usage

BestConst(y)

Arguments

y

outcomes, non-empty.

Value

list(best_prediction, mse).

References

Morin, D. J. (2016). Probability: For the Enthusiastic Beginner. Createspace. Eqs (6.22)-(6.23).

Examples

BestConst(c(1, 1, 3, 4, 6))$best_prediction
#> [1] 3