mice
mice can finally predict, not just estimate, from multiply imputed data.
A side-by-side editorial comparison of midr and Nextflow — release velocity, themes, recent moves, and the top alternatives to consider.
A black-box interpreter reaches CRAN, then learns multi-class and survival responses
midr explains black-box models by fitting an interpretable surrogate through Maximum Interpretation Decomposition — main effects plus second-order interactions, with exact Shapley values for the surrogate. Two months after its first CRAN release it can take a matrix response, which covers multi-class classification and survival models, and hold collections of fitted interpretations in midlist and midrib objects for comparison.
Nextflow just made AI agents a task type, alongside processes and containers.
Nextflow ships on two tracks — dated edge releases carrying new work and 25.10.x/26.04.x stable lines taking backports — and the split is visible in every window: the edge builds run to dozens of commits while the stable patches are often plugin bumps alone. The current edge, 26.08.0-edge, introduces an agent primitive that makes AI agents first-class tasks, on top of steady work hardening the v2 config parser and type system. Cloud-executor plumbing, particularly for Seqera's own scheduler, is the other constant.
midr explains black-box models by fitting an interpretable surrogate through Maximum Interpretation Decomposition — main effects plus second-order interactions, with exact Shapley values for the surrogate. Two months after its first CRAN release it can take a matrix response, which covers multi-class classification and survival models, and hold collections of fitted interpretations in midlist and midrib objects for comparison.
The releases move outward along two axes at once: what can be interpreted, and how much of it fits in memory. Version 0.5.3 rebuilt the fitting path to avoid materialising large design matrices and added a save.memory option; 0.6.0 widened the response from a vector to a matrix and added parametric link functions. Class and argument names were shortened in the same release, so the package is still willing to break itself this early.
With multiple models now held in one object and visualisation methods for them, comparison across models is the surface most likely to fill out next — the collection classes exist but the notes describe manipulation and plotting rather than any comparison metric.
Nextflow ships on two tracks — dated edge releases carrying new work and 25.10.x/26.04.x stable lines taking backports — and the split is visible in every window: the edge builds run to dozens of commits while the stable patches are often plugin bumps alone. The current edge, 26.08.0-edge, introduces an agent primitive that makes AI agents first-class tasks, on top of steady work hardening the v2 config parser and type system. Cloud-executor plumbing, particularly for Seqera's own scheduler, is the other constant.
Two arcs run in parallel. One is language maturity: the v2 parser, record types, typed outputs and the formatter have absorbed fix after fix across every release here, which is what a language does before it declares a syntax stable. The other is consolidation around Seqera — nf-tower folded into nf-seqera at 1.0.0, scheduler run identifiers propagate to Platform, and per-user CPU caps and secret references now cross that boundary. The agent primitive is the newest and least settled direction, arriving in edge rather than a stable line.
The agent primitive lands in an edge build, so the near-term question is whether it survives into a stable line unchanged or is reshaped first; the 26.04.x and 25.10.x branches show no sign of it yet. Expect the next edge releases to iterate on its interface while the stable branches continue taking parser and executor backports.
Other Infra & APIs products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either midr or Nextflow.
mice can finally predict, not just estimate, from multiply imputed data.
A market-microstructure toolkit that keeps adding estimators as the papers land.
A vowel-analysis package trimming dependencies after an email address got it archived.
The R half of the EMU speech database system, fixing what was quietly broken.
A Bayesian model-averaging package spending its 2.0 on memory, not methods.
tidyplots keeps rebuilding its own foundations rather than layering around them.
See all midr alternatives → · See all Nextflow alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Nextflow is currently shipping more aggressively (velocity 6.3 vs 0.0), with 1 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. Nextflow is currently shipping more aggressively (velocity 6.3 vs 0.0), with 1 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Infra & APIs products to evaluate alongside.
Top midr alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "midr alternatives" section above for the current picks, or visit /alternatives/midr for the full list with editorial commentary on each.
Top Nextflow alternatives in Infra & APIs are ranked by recent ship velocity. Browse the "Nextflow alternatives" section above for the current picks, or visit /alternatives/nextflow for the full list with editorial commentary on each.