-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathissues.jsonl
More file actions
20 lines (20 loc) · 12.8 KB
/
Copy pathissues.jsonl
File metadata and controls
20 lines (20 loc) · 12.8 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
{"id":"referent-c48","title":"Dynamics kernel estimator: scale-unidentified, silent fallback to start values","description":"R/dynamics.R:104-151. process_correlation normalises so (tau_b,tau_g,sigma_e) are unidentified; BFGS walks ridge, eigen fails, tryCatch returns START values with identified=TRUE. norm_stable (ell=Inf) never fits. Fix: simplex parameterization + bounded log(ell), L-BFGS-B, check convergence, never silent fallback, stable+nugget when lags fixed, report normalised variance fractions. Also A4: NA velocity/forecast quantities when unidentified. Add OOS test on irregular lags: var(innovation) in 1±0.1, FPR in 0.05±0.015. Review A3/A4.","status":"closed","priority":0,"issue_type":"bug","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:18Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:28:20Z","started_at":"2026-08-22T13:02:50Z","closed_at":"2026-08-22T13:28:20Z","close_reason":"Phases 2-3","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-0rh","title":"Fix site scale adaptation bias (E[log|Z|] = -0.635)","description":"R/adapt.R:47-50 uses mean(log|r/s|); identical-site adaptation shrinks sigma to 0.55x, held-out var(z)=3.2-3.9. Fix: 0.5*log(mean((r/s)^2)) with z=qnorm(centile); add test that identical-site adaptation leaves var(z)~1. Review A2.","status":"closed","priority":0,"issue_type":"bug","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:17Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:28:19Z","started_at":"2026-08-22T13:02:49Z","closed_at":"2026-08-22T13:28:19Z","close_reason":"Phases 2-3","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-4du","title":"Derive covariates from spec formulas, not all non-outcome columns","description":"R/fit.R:30,45,260. Covariates are every non-outcome column: NA in unused columns silently drops rows; Mahalanobis support is computed over other outcomes; predict errors if unused column absent. Fix: all.vars() over spec formulas; complete-case on outcome+those; message dropped rows. See docs/review-2026-08-22.md A1.","status":"closed","priority":0,"issue_type":"bug","assignee":"bbuchsbaum","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:15Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:02:48Z","started_at":"2026-08-22T12:51:42Z","closed_at":"2026-08-22T13:02:48Z","close_reason":"Phase 1 fdb58e2","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-9ra","title":"Cut dead/half-baked surface: gamlss engine, ordinal/discrete, parallel.R, as_wide, filter_scores, dead utils; merge velocity/stable/card; internalize scale_*/fortify_*","description":"See docs/review-2026-08-22.md section B. Drop Matrix and scales Imports. Wire future.apply into norm_fit directly.","status":"closed","priority":1,"issue_type":"task","assignee":"bbuchsbaum","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:24Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:02:48Z","started_at":"2026-08-22T12:51:42Z","closed_at":"2026-08-22T13:02:48Z","close_reason":"Phase 1 fdb58e2","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-eiv","title":"NA predictors error on location-scale fits; seed/determinism for uncertainty='total'","description":"engine-mgcv.R:121, dist.R:44 (A8); engine-mgcv.R:616 unseeded 64 draws, cdf() ignores draws (A9). Also non-syntactic outcome names (A11).","status":"closed","priority":1,"issue_type":"bug","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:23Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:28:20Z","started_at":"2026-08-22T13:02:50Z","closed_at":"2026-08-22T13:28:20Z","close_reason":"Phases 2-3","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-aoa","title":"Tail/boundary handling: log-space tails, no 1e-12 clamp, ecdf boundaries, joint n_observed","description":"calibrate.R:39-42, joint.R:25-41, utils.R:5, dist.R:247. Review A7.","status":"closed","priority":1,"issue_type":"bug","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:21Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:28:20Z","started_at":"2026-08-22T13:02:49Z","closed_at":"2026-08-22T13:28:20Z","close_reason":"Phases 2-3","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-068","title":"Collapse norm_calibrated/norm_adaptation/norm_reference class stacking into predict.norm_fit","description":"predict.norm_adaptation ignores allow_extrapolation, drops total draws and .id, never applies calibration, returns plain tibble; norm_reference drops norm_calibrated class. One norm_fit with optional $calibration/$adaptation slots and a shared scores_from_dists(). Review A6.","status":"closed","priority":1,"issue_type":"bug","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:20Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:28:19Z","started_at":"2026-08-22T13:02:49Z","closed_at":"2026-08-22T13:28:19Z","close_reason":"Phases 2-3","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-6ca","title":"norm_select calibration gate and CRPS are in-sample","description":"R/select.R:50-59,107: split_for_assess is a stub. Compute gates from crossfit scores; paired-difference SE for the 1-SE rule; actually use CRPS. Review A5.","status":"closed","priority":1,"issue_type":"bug","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:19Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T13:28:19Z","started_at":"2026-08-22T13:02:49Z","closed_at":"2026-08-22T13:28:19Z","close_reason":"Phases 2-3","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-ykj","title":"First-class ggplot2 plotting layer","description":"Replace the autoplot scaffold with a publication theme, named centile charts (optional grouping), composition diagnostics, forecast/thrive fans, and trajectory overlays. ggplot2 + scales; patchwork/ggridges only if Suggests.","status":"closed","priority":1,"issue_type":"feature","assignee":"bbuchsbaum","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:15:39Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T12:20:44Z","started_at":"2026-08-22T12:16:08Z","closed_at":"2026-08-22T12:20:44Z","close_reason":"First-class ggplot2 layer: theme_referent, fortify helpers, named centile charts, composition, forecast fans, trajectory overlays.","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-w0b","title":"Expand growth vignette into a real walkthrough","description":"Fit, assess, autoplot, score, flag/filter. Knit and fix API friction (id on scores, autoplot export).","status":"closed","priority":1,"issue_type":"task","assignee":"bbuchsbaum","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T03:12:44Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T03:15:30Z","started_at":"2026-08-22T03:12:51Z","closed_at":"2026-08-22T03:15:30Z","close_reason":"Growth vignette walks fit/assess/autoplot/score/flag. Id is excluded from covariates; autoplot is re-exported.","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-5cv","title":"Change-score FPR stays near nominal","description":"Sweep residual correlation, lag, and measurement noise. norm_transition() |change_z|\u003e1.96 stays near 5%; naive Z2-Z1 does not.","status":"closed","priority":1,"issue_type":"task","assignee":"bbuchsbaum","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T03:12:43Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T03:15:29Z","started_at":"2026-08-22T03:12:50Z","closed_at":"2026-08-22T03:15:29Z","close_reason":"Analytic and package-level FPR sweeps: change_z stays near 5% across r/lag/noise; naive Z2-Z1 does not.","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-au6","title":"SHASH should beat Gaussian on skewed OOS scores","description":"Add a statistical gate: on strongly skewed data, SHASH has better held-out log score, more uniform PIT, and flatter worm residuals than a Gaussian location-scale model.","status":"closed","priority":1,"issue_type":"task","assignee":"bbuchsbaum","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T03:12:42Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T03:15:28Z","started_at":"2026-08-22T03:12:49Z","closed_at":"2026-08-22T03:15:28Z","close_reason":"SHASH beats Gaussian on skewed OOS log score, PIT, and worm residuals; Gaussian stays competitive on Gaussian data.","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-xei","title":"Phase 1: scientific kernel","description":"Distribution contract, mgcv engines, scores, assess, crossfit, support, simulation laws.","status":"closed","priority":1,"issue_type":"epic","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-21T23:23:28Z","created_by":"bbuchsbaum","updated_at":"2026-08-21T23:23:33Z","closed_at":"2026-08-21T23:23:33Z","close_reason":"Implemented in the initial package kernel, usable layer, and extensions.","dependency_count":0,"dependent_count":1,"comment_count":0}
{"id":"referent-0th","title":"Vignettes: brain-charts-across-sites, longitudinal, coming-from-PCNtoolkit; replace stubs","description":"Review F.","status":"in_progress","priority":2,"issue_type":"task","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:27Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T14:38:54Z","started_at":"2026-08-22T14:38:54Z","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-dsz","title":"Strengthen tautological tests; cover norm_derivative, variance, predict.norm_reference, bam path","description":"Review E.","status":"in_progress","priority":2,"issue_type":"task","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:26Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T14:38:54Z","started_at":"2026-08-22T14:38:54Z","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-m7u","title":"Performance: strip gam objects on freeze, consolidate predict.gam calls, vectorise calibration/CRPS","description":"Review D. norm_reference currently larger than the fit; predict.gam called 2-3x per outcome; apply_calibration per-row ecdf; crps per-row vec_slice.","status":"in_progress","priority":2,"issue_type":"task","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:25Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T14:38:54Z","started_at":"2026-08-22T14:38:54Z","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-ef3","title":"Migrate norm_dist to the distributional package; add tidy/glance/augment via generics","description":"Review C1-C2. dist_shash + dist_conditioned as distributional custom dists; dist_mixture for total uncertainty (removes cdf_total); augment() returns wide Z matrix. Add ev/smse metrics.","status":"closed","priority":2,"issue_type":"feature","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:24Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T14:38:53Z","started_at":"2026-08-22T13:28:22Z","closed_at":"2026-08-22T14:38:53Z","close_reason":"Phase 4 86b94b8","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-20t","title":"Phase 2: usable reference-model package","description":"Graphics, adaptation, recalibration, model cards, parallelism, vignettes.","status":"closed","priority":2,"issue_type":"epic","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-21T23:23:30Z","created_by":"bbuchsbaum","updated_at":"2026-08-21T23:23:33Z","closed_at":"2026-08-21T23:23:33Z","close_reason":"Implemented in the initial package kernel, usable layer, and extensions.","dependencies":[{"issue_id":"referent-20t","depends_on_id":"referent-xei","type":"blocks","created_at":"2026-08-21T19:23:31Z","created_by":"bbuchsbaum","metadata":"{}"}],"dependency_count":1,"dependent_count":1,"comment_count":0}
{"id":"referent-0go","title":"CRAN hygiene + plots: non-ASCII en dash in plot-data.R, missing \\value, set.seed in package code, cli vs cat, placeholder plots","description":"Review F.","status":"in_progress","priority":3,"issue_type":"task","assignee":"bbuchsbaum","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-22T12:45:28Z","created_by":"bbuchsbaum","updated_at":"2026-08-22T14:38:55Z","started_at":"2026-08-22T12:51:42Z","dependency_count":0,"dependent_count":0,"comment_count":0}
{"id":"referent-013","title":"Phase 3: restrained extensions","description":"Dynamics, joint copula, discrete PIT, optional gamlss engines.","status":"closed","priority":3,"issue_type":"epic","owner":"brad.buchsbaum@gmail.com","created_at":"2026-08-21T23:23:30Z","created_by":"bbuchsbaum","updated_at":"2026-08-21T23:23:33Z","closed_at":"2026-08-21T23:23:33Z","close_reason":"Implemented in the initial package kernel, usable layer, and extensions.","dependencies":[{"issue_id":"referent-013","depends_on_id":"referent-20t","type":"blocks","created_at":"2026-08-21T19:23:31Z","created_by":"bbuchsbaum","metadata":"{}"}],"dependency_count":1,"dependent_count":0,"comment_count":0}