Help us improve
Share bugs, ideas, or general feedback.
Identifies statistical outliers and contextual anomalies in datasets, flagging unusual values and potential stories for journalists and data investigators.
npx claudepluginhub ur-grue/autopunk-media-skills --plugin autopunk-media-skillsHow this skill is triggered — by the user, by Claude, or both
Slash command
/autopunk-media-skills:data-outlier-finderThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Identifies unusual values, unexpected patterns, and potential stories hidden in a dataset by systematically checking for statistical outliers and contextual anomalies.
Finds newsworthy angles, outliers, trends, and comparisons hidden in datasets. Use when you need to pitch a data story or stress-test a dataset before reporting.
Analyze preprocessed data for investigative journalism with full transparency. Use when a journalist has clean, preprocessed data ready for analysis and needs to identify patterns, anomalies, relationships, or statistical findings that support a story. Triggers include requests to analyze data, find patterns, identify outliers, cross-reference records, calculate statistics, or answer specific investigative questions. Complements the structured-data-preprocessing skill. Emphasizes simple, legible analyses over complex methods—every finding must be explainable to editors and defensible under scrutiny.
Applies descriptive stats, trend analysis, outlier detection, hypothesis testing to distributions, anomalies, correlations, and business metrics.
Share bugs, ideas, or general feedback.
Identifies unusual values, unexpected patterns, and potential stories hidden in a dataset by systematically checking for statistical outliers and contextual anomalies.
Required:
Optional:
A numbered list of findings, ordered by editorial potential. Each finding includes:
Ends with a short summary: how many outliers were found, how many are likely noise vs. potential stories, and what additional data would help confirm the findings.
Dataset: Annual per-capita healthcare spending (USD, PPP-adjusted) for 30 OECD countries, 2018–2023. Source: OECD Health Statistics. I want to find anything surprising that could anchor a comparative healthcare story for a general-interest magazine.
Finding 1: The United States remains a dramatic outlier at $12,555 per capita (2023) The US spends 2.3× the OECD median ($5,460). This gap has widened, not narrowed, since 2018 (+18% vs. OECD median +12%). The editorial angle isn't the outlier itself — that's well known — but the acceleration. The gap is growing faster than most readers assume. Reporting question: What specific cost categories are driving the widening gap? Is it pharmaceuticals, administrative overhead, or specialist fees?
Finding 2: Turkey's spending jumped 34% between 2021 and 2023 From $1,650 to $2,210 — the fastest growth rate in the dataset by a wide margin. The OECD average grew 8% over the same period. This is not a statistical artefact; the increase is consistent across both years. Reporting question: Is this driven by post-earthquake health infrastructure spending, a policy expansion, or currency/PPP adjustment effects? Check whether the increase appears in nominal local-currency terms or only in PPP-adjusted dollars.
Finding 3: Japan's spending barely moved (+2%) despite an aging population Japan spent $4,691 in 2023, up from $4,587 in 2018. In a country where 29% of the population is over 65, flat healthcare spending is the outlier — it contradicts the standard assumption that aging populations drive cost increases. Reporting question: Is Japan suppressing costs through price controls, care rationing, or efficiency gains? Or is the flat figure masking a shift from public to out-of-pocket spending?
Finding 4: Luxembourg appears as a statistical outlier at $7,890, but it's probably noise Luxembourg's small population (660,000) and high cross-border workforce distort per-capita calculations. Many people receiving care in Luxembourg aren't counted in the population denominator. This is a known measurement issue in OECD data. Reporting question: None — flag this as a data limitation rather than a story.
Summary: 4 outliers identified in 180 data points (30 countries × 6 years). Finding 1 is well-trodden but the acceleration angle is fresh. Finding 2 is the strongest lead for original reporting — under-covered and verifiable. Finding 3 is a strong counterintuitive angle for a feature. Finding 4 is noise. Additional data that would help: spending broken down by cost category (hospitals, pharma, admin) and out-of-pocket vs. public expenditure splits.