AGENT SKILL

sweetviz

Use Sweetviz to generate pandas EDA reports, compare tabular datasets, configure feature typing, and troubleshoot HTML/notebook output.

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

01 前言 Front Matter

name
sweetviz
license
MIT
metadata
{ "disco-role": "operating" }
description
Use Sweetviz to generate pandas EDA reports, compare tabular datasets, configure feature typing, and troubleshoot HTML/notebook output.
disable-model-invocation
true

02 资源文件 5

03 SKILL.md 正文

Sweetviz repo skill

Use this skill when a task involves Sweetviz, pandas exploratory data analysis reports, target-aware tabular profiling, train/test dataset comparison, feature type overrides, Sweetviz config files, or Sweetviz HTML/notebook output troubleshooting.

Quick start

  1. Install Sweetviz in the active project environment:

    python -m pip install sweetviz
    

    For source checkouts or local fixes, use an editable install from that checkout:

    python -m pip install -e .
    
  2. Verify the public import and signatures before writing guidance that depends on exact parameters:

    import sweetviz as sv
    print(sv.__version__)
    print(sv.analyze)
    

    Or run the bundled environment checker:

    python scripts/check_sweetviz_install.py --json
    
  3. Route to a sub-skill for the actual workflow.

Route map

  • Use sub-skills/report-generation/ when the user wants to create a Sweetviz report from one pandas DataFrame, choose a target, save HTML with show_html(), embed with show_notebook(), control layout/scale, avoid browser launches, or understand optional Comet logging.
  • Use sub-skills/dataset-comparison/ when the user wants to compare train/test DataFrames, baseline/current datasets, or two groups from one DataFrame with compare() or compare_intra().
  • Use sub-skills/configuration-and-data-handling/ when the user mentions FeatureConfig, skipped/forced feature types, wrong type inference, config overrides, quiet/progress-only output, duplicate columns, target NaNs, mixed types, CJK fonts, or preflight validation.

Common tasks often need two routes: validate/force feature types with configuration-and-data-handling, then render with report-generation or compare with dataset-comparison.

Public API orientation

Sweetviz 2.3.3 exposes these primary objects from import sweetviz as sv:

  • sv.analyze(source, target_feat=None, feat_cfg=None, pairwise_analysis='auto')
  • sv.compare(source, compare, target_feat=None, feat_cfg=None, pairwise_analysis='auto')
  • sv.compare_intra(source_df, condition_series, names, target_feat=None, feat_cfg=None, pairwise_analysis='auto')
  • sv.FeatureConfig(skip=None, force_cat=None, force_text=None, force_num=None)
  • sv.DataframeReport with show_html(), show_notebook(), and log_comet() methods
  • sv.config_parser for INI-style defaults such as verbosity, layouts, CJK fonts, and type thresholds

Do not pass verbosity= to public sv.analyze() or sv.compare() in this version. Use [General] default_verbosity through sv.config_parser or the lower-level DataframeReport(..., verbosity=...) path when that trade-off is intentional.

Shared references and scripts

  • Read compatibility and assets for package dependencies, Python support, optional extras, packaged templates/fonts/styles, and why there is no Sweetviz CLI route.
  • Read troubleshooting for install/import failures, shadowed modules, package-data/template errors, font/browser/notebook issues, optional Comet behavior, and broad data compatibility symptoms.
  • Read repo provenance before deciding whether this skill is current for a Sweetviz checkout or should be refreshed.
  • Run check_sweetviz_install.py to verify importability, version metadata, public signatures, packaged assets, and optional tiny HTML rendering without opening a browser.

Cautions

  • Targets must exist in the source data, must not be skipped, must contain no missing values, and must be boolean or numeric after Sweetviz type detection/forcing.
  • Numeric columns with at most the configured distinct-value threshold infer as categorical by default. Force known numeric low-cardinality columns with FeatureConfig(force_num=[...]) when needed.
  • pairwise_analysis='auto' can warn and return early for wide data; use 'off' for deterministic smoke reports or 'on' only when the user explicitly wants associations and accepts the cost.
  • Sweetviz has no console entry point in this package version. Use the Python API or bundled skill helper scripts.
  • Optional Comet.ml logging is external and credentialed. Do not require it for local report generation or verification unless the user explicitly asks for Comet integration.

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