preml: Compute 1- or 2-way pseudo-REML estimates

In damian-t-p/halfsibdesign: What the Package Does (One Line, Title Case)

Compute 1- or 2-way pseudo-REML estimates

Description

Given either a tuple of sum-of-squares matrices or a dataframe, computes preudo-REML estimates for the covariances.

Usage

preml_1way(Mw, r, Mb, n)

preml_2way_mat(M1, I1, M2, I2, M3, I3)

preml_2way_df(df, ...)

Arguments

Mw, Mb	Sum-of-squares within and between matrices.
r, n	Number individuals per group and groups.
M1, M2, M3	Sum-of-squares matrices within sire, dam and individual groups.
I1, I2, I3	Number of sires, dams per sire and individuals per dam.
df	Data frame for 2-way design.
...	Other arguments.

Details

The pseudo-REML estimates are non-negative definite estimates of the relevant covariance matrices as described in [1]. These are not the exact REML estimates, but coincide with them in the MANOVA estimates are non-negative definite.

Value

A list with entries

• in the one-way case: Sb, Sw, which are the estimates of the between-groups and within-groups covariance matrices.

• in the two-way case: S1, S2 and S3, which are the estimates of the sires, dams and individual covariance matrices.

References

[1] Y. Amemiyah. "What Should be Done When an Estimated between-Group Covariance Matrix is not Nonnegative Definite?" In The Americal Statistician 39.2 (1985).

See Also

preml_2way for generic version.

damian-t-p/halfsibdesign documentation built on March 14, 2023, 4:55 a.m.