New Features and Changes in PACE

What's New in release 2.17?

1) A new package, PACE-kcfc that is used for clustering of functional data.

References: Chiou, J.M. and Li, P.L. (2007). Functional clustering and identifying substructures of longitudinal data[J]. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 69(4): 679-699.

2) A new method for estimating the eigenvalues under the sparse case (regular=0) is added (option ls_fit=1). This method uses the eigenfunctions obtained by smoothing the covariance matrix, and then uses least squares to regress the raw covariance on the eigenfunctions.

3) For dense regular functional data with number of observed time points much larger than sample size, there is a new method for estimating mean and covariance function, eigenvalues and eigenfunctions by specifying regular = 3. This method assumes no measurement error.

4) Some minor bugs are fixed for subroutines PACE-mani and PACE-GFLM.

5) All 'pinv' in base PACE have been replaced by either 'mldivide' or 'mrdivide' for enhanced efficiency. (Files involved: PCA.m, rotate_mlwls.m, FPCApred.m, getLogLik1.m, getLogLik2.m, getOriCurves.m, getScores1.m, getScores2.m, lwls.m, mullwlsk.m)

6) Minor color issue in designPlot() is fixed.

Click here to download the full change logs for all versions.