NEWS


Stem 2.0.0 (2026-11-21)

First release of the package after its archival on CRAN, and the first to include the spatially-clustered STEM (SC-STEM) model family. The classical STEM workflow of version 1.0 is unchanged in its statistical content; what changes is its numerical robustness, the whole SC-STEM layer, and the packaging.

Breaking changes: names aligned with the paper and with SCDA

New features

Regularized regression coefficients

Spatially-clustered STEM models

Missing values in the response, fitted values

Algorithmic changes in the SC-STEM assignment step

Corrections to the SC-STEM assignment score

The score used to allocate locations to clusters before 2.0.0 contained four defects, all fixed:

The contribution is now an explicit conditional pseudo-likelihood, documented as such, and used only to rank clusters: coefficients, variance components and information criteria all come from the exact cluster-wise likelihoods.

Information criteria

The log-likelihood returned by STEM_Estimation()

Convergence of the EM algorithm and computational settings

Numerical robustness of the STEM core

These were pre-existing defects, harmless for a single pooled fit but fatal for a clusterwise algorithm, which refits the model on hundreds of different subsets of locations. Both aborted the fit with missing value where TRUE/FALSE needed:

On the pm10 example these fixes take the parametric bootstrap from 4 usable draws out of 12 to 12 out of 12.

Larger networks

CRAN compliance

Addresses the review comments received on the previous submission:

Further packaging work required before submission:

Documentation

Stem 1.0 (2012-10-29)