메뉴
Apply statistical methods including descriptive stats, trend analysis, outlier detection, and hypothesis testing. Use when analyzing distributions, testing for significance, detecting anomalies, computing correlations, or interpreting statistical results.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
新しいエージェントを雇用（作成・起動）するスキル。ディレクトリ作成、CLAUDE.md生成、スキルコピー、roster更新、tmux起動まで一貫して行う。
Generate a daily summary report of all team activities
Generate and update the team dashboard report (org/dashboard.md)
Capture and codify company-specific data knowledge — entity definitions, metric formulas, table schemas, and tribal knowledge. Use when onboarding to a new data environment, documenting institutional data knowledge, or building a data dictionary.
Profile and explore datasets to understand their shape, quality, and patterns before analysis. Use when encountering a new dataset, assessing data quality, discovering column distributions, identifying nulls and outliers, or deciding which dimensions to analyze.
QA an analysis before sharing with stakeholders — methodology checks, accuracy verification, and bias detection. Use when reviewing an analysis for errors, checking for survivorship bias, validating aggregation logic, or preparing documentation for reproducibility.
| name | statistical-analysis |
| description | Apply statistical methods including descriptive stats, trend analysis, outlier detection, and hypothesis testing. Use when analyzing distributions, testing for significance, detecting anomalies, computing correlations, or interpreting statistical results. |
Descriptive statistics, trend analysis, outlier detection, hypothesis testing, and guidance on when to be cautious about statistical claims.
| Situation | Use | Why |
|---|---|---|
| Symmetric distribution, no outliers | Mean | Most efficient estimator |
| Skewed distribution | Median | Robust to outliers |
| Categorical or ordinal data | Mode | Only option for non-numeric |
| Highly skewed with outliers | Median + mean | Gap shows skew |
Always report mean and median together for business metrics.
p1 (floor), p5 (low normal), p25 (Q1), p50 (median), p75 (Q3), p90 (power users), p95 (high normal), p99 (extreme)
Always communicate uncertainty as ranges, not point estimates.
Do NOT automatically remove. Instead:
A/B test results, before/after comparisons, segment comparisons.
| Scenario | Test |
|---|---|
| Compare two group means | t-test (independent) |
| Compare two proportions | z-test for proportions |
| Paired measurements | Paired t-test |
| Compare 3+ group means | ANOVA |
| Non-normal data | Mann-Whitney U test |
| Categorical association | Chi-squared test |
A difference can be statistically significant but practically meaningless. Always report:
Consider reverse causation, confounding variables, and coincidence.
Testing 20 metrics at p=0.05 means ~1 will be falsely significant. Adjust with Bonferroni or report number of tests.
Aggregated trends can reverse when segmented. Always check key segments.
Analyzing only survivors ignores churned/failed entities. Ask "who is missing?"
Use ranges, not exact numbers. "About 5%" is more honest than "4.73%".
