Estimate missing landmark data
This function provides several options for estimating landmark data (details of which can be found in the references below). The function first alignes the landmarks using Procrustes superimposition (
align.missing). Both 2D and 3D coordinates can be accommodated.
MissingGeoMorph(x, nlandmarks, method = "BPCA")
A n* l X 2 matrix of coordinate data, where n is the number of specimens and l is the number of landmarks. All landmarks from one specimen should be grouped together. Missing values should be given as NA
The number of landmarks per specimen.
Four methods are provided for estimating missing landmark data: 1) "BPCA" - Bayesian principal component analysis, 2) "mean" - mean substitution, 3) "reg" - values are estimated based on the most strongly correlated variable available, and 4) "TPS" - thin plate spline interpolation (only available for 2D). See Arbour and Brown (2014) for a comparison of the performance of each of these methods.
Returns an n * l X 2 (or 3) matrix of coordinate data, with missing values imputed. Landmarks have been aligned and are given in the original shape space.
Arbour, J. and Brown, C. 2014. Incomplete specimens in Geometric Morphometric Analyses. Methods in Ecology and Evolution 5(1):16-26.
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.