AGENT SKILL

data-processing-and-covariates

Use Darts preprocessing transformers, Pipeline, missing-value filling, scaling, inverse transforms, and covariate span checks.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
data-processing-and-covariates
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Use Darts preprocessing transformers, Pipeline, missing-value filling, scaling, inverse transforms, and covariate span checks.
disable-model-invocation
true

02 资源文件 4

03 SKILL.md 正文

Data processing and covariates

Use this sub-skill when the user needs to clean/transform Darts TimeSeries, fit preprocessing only on training data, inverse-transform forecasts, or build/validate past and future covariates.

Read first

Route by task

  • Fill missing target values: use MissingValuesFiller, usually inside a Pipeline.
  • Scale data: fit Scaler or a Pipeline on train only, then transform validation/test; never fit on held-out data.
  • Invert forecast scale: keep the fitted transformer and call inverse_transform() on the forecast with matching components.
  • Generate calendar covariates: use Darts time-series generation helpers such as datetime/day-of-week attributes or caller-provided covariate tables, then stack components.
  • Fix covariate span errors: validate target end time, forecast horizon, covariate start/end, frequency, and one covariate series per target series.
  • Choose a model that consumes covariates: route to ../forecasting-workflows/ for core/regression models or ../torch-and-foundation-models/ for neural models.

Safe check

python scripts/transform_pipeline_smoke.py --quiet

The script generates a small daily series, fills/scales train data, transforms validation data, inverse-transforms a toy forecast, stacks two future covariates, and asserts horizon coverage.

Boundaries

This sub-skill prepares data and covariates. It does not decide all model-family details or interpret final metrics. Do not solve future covariate span errors by extending the target series with fake values; extend/regenerate the covariate series instead.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351