菜单
Writes findings in language the evidence can carry — the claim ladder (observed → directional → decision-support, never causal without a design), effect sizes over drama, banned phrasings. Use when writing any conclusion, report section, playbook, or public page from analysis results.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Finds label leakage and designs honest validation — leakage taxonomy, grouped and time-aware splits, base rates, and the attack-your-own-model checklist. Use before trusting any metric, when a score looks too good, or when features and labels share time windows or origins.
The FlyRank internship datasets — the 30k-row starter CSV and its gotchas, the ~79M-row warehouse release tables and grains, panel warnings, access, and iteration rules. Load for EVERY task that touches the data. (Project-specific: delete this folder when reusing the skill library elsewhere.)
Works with datasets far too big to download or load in pandas — SQL over remote Parquet with DuckDB, aggregate-then-model, iterate on samples. Use when a dataset has millions of rows, lives on a remote host (hf:// or s3), or a notebook runs out of memory or hits rate limits.
Runs an honest EDA and signal audit — distributions first, heavy-tail handling, verdict-style mini-tests (CONFIRMED/OPPOSITE/MIXED/FALSE), sample-size floors. Use before modeling, when testing whether a popular belief shows up in data, or when a correlation looks too good.
Builds the transparent rule-based baseline every model must beat — a hand-written score with reason codes, ranked output, and precision@K evaluation. Use before training any model, or when someone reports model results with nothing to compare against.
Deploys a static page (research paper, portfolio piece) for free from a GitHub repo using GitHub Pages — setup, file layout, verification, and recording the final URL. Use when publishing a finished write-up as a public web page.
| name | writing-honest-claims |
| description | Writes findings in language the evidence can carry — the claim ladder (observed → directional → decision-support, never causal without a design), effect sizes over drama, banned phrasings. Use when writing any conclusion, report section, playbook, or public page from analysis results. |
A finding is a sentence plus the evidence that carries it. Most analysis goes wrong in the sentence, not the math: the words claim more than the numbers showed.
| Evidence you actually have | Words you may use |
|---|---|
| A pattern in this dataset, this period | "we observed…", "in this data…" |
| A measured comparison between groups | "X is associated with Y", "pages with A showed B" |
| A validated model that ranks/predicts out-of-sample | "the model ranks/flags… at precision@K of…" |
| A controlled experiment or matched design | only THEN: "X causes/improves Y" |
Cross-sectional data (one snapshot, no intervention) NEVER supports "doing X will produce Y." The honest form is decision-support: "these pages look worth reviewing first, because…"
Ask three questions of every bold sentence: Where does the label come from? What does the validation design actually test? Would the number survive a grouped/time split? Frame findings as "how to make it stronger", not "gotcha" — the goal is the next level of rigor, not a scalp.
