Hacktoberfest 2026: le issue che i maintainer hanno segnato per ottobre, aperte e adatte ai principianti. Sfoglia le issue Hacktoberfest

Extend partner-dependence checks to intensive longitudinal data

Aperta
  • Qualcosa non va?
#24 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

@Pascal-Kueng ci sta già lavorando.

Dal 29/9/2026.

  • #81 di @Pascal-Kueng — aperta

Valutazione

Difficoltà
5/5
Tempo stimato
Più di una settimana
Idoneità per principianti
30/100
Tipo di issue
Funzionalità
Chiarezza
Specificata chiaramente
Stato di attività
Tranquilla
Stack tecnologico
r
Ambito
testing

Direzione di ricerca

Start by reading issue #22 and the existing Gaussian prototype branch, and check whether #18 and the cross-sectional PR have merged — this issue is blocked on both. Then create branch generalized-ild-checks from updated main, using DHARMa::testGeneric() as the reference for formal tests. Done means a longitudinal PR where Gaussian, Poisson, NB1/NB2, Tweedie, Gamma and beta each pass validation on raw and model-centred outcomes, results match the prototype, and full package checks pass.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

enhancement priority: medium

Extend partner-dependence checks to intensive longitudinal data (ILD): repeated measurements from each dyad. Reuse #22 after #18 and the cross-sectional PR below are merged. Validate each outcome distribution before claiming support.

Keep the visual checks and existing descriptive summaries. Zero-inflated and hurdle models, differences between observed and simulated summaries, ratios of standard deviations or root mean squares, and p-values are out of scope. For later formal tests, use DHARMa::testGeneric() as a reference and validate them before claiming support.

Cross-sectional PR (one occasion per dyad)

  • Merge #18.
  • Rebase simple-generalized-checks onto the updated main.
  • Open the PR for cross-sectional checks across supported model families.
  • Review the PR.
  • Merge the PR.

Longitudinal PR

  • Create generalized-ild-checks from the updated main.
  • Bring over the longitudinal checks from #22. Keep the Gaussian prototype unchanged until its results have been compared with the new version.
  • Apply the same rules for unavailable correlations:
    • Keep rows for requested lags that the data do not support, and explain why.
    • Give an error if an observed correlation should be calculable but is not.
    • Give an error if no simulation produces a calculable correlation.
    • If only some simulations cannot produce a correlation, warn once and use the rest.
  • Support Gaussian, Poisson, negative binomial (NB1 and NB2), Tweedie, Gamma, and beta models only after each family passes validation.
  • Confirm that all calculable Gaussian results match the prototype, apart from the correlation rules above.
  • Confirm that cross-sectional checks remain unchanged.
  • Validate every supported family on original outcomes (raw) and after subtracting model predictions (model-centred). For the latter, subtract the same prediction from each observed value and its simulations.
  • Test structured random effects such as AR(1), predictor-dependent dispersion (dispformula), sparse outcomes such as many zeros, missing measurements, unequal measurement counts across people or dyads, and correlations unavailable in some simulations.
  • Add a small simulation study comparing models with adequate dependence structures with models missing relevant dependence. Briefly summarize the validation results.
  • Run full package checks and update the documentation.
  • Open the longitudinal PR.
  • Close #22 once the new PR clearly replaces it.
  • Review the new PR.
  • Merge the new PR.
  • Delete the Gaussian prototype branch after the new PR is merged.

Initial Poisson AR(1) and NB2 checks show the approach is feasible; they do not replace validation for each family.

Lingua principale
R
Stelle
1
Fork
2
Merge medio
1g 2h
PR unite (30g)
26

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di Pascal-Kueng/dyadMLM

Tutte le issue di Pascal-Kueng/dyadMLM

Issue simili

Altre issue su R

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.