Aller au contenu

Données ouvertes

Bibliothèque

Jeux de données ouverts, entièrement documentés — interrogeables ici, et lisibles par n’importe quel LLM.

Les titres et les descriptions proviennent des sources de données, en anglais.

153 jeux de données

signals

  • Inflation Signals (derived)

    US inflation signals (CPI/PCE YoY momentum, core gaps, anomalies, forecasts)

    Signals derived from FRED's US inflation series: 30-period annualized volatility of monthly changes, 3-month momentum, year-over-year percent change (the headline inflation gauge), 3-sigma anomaly flags, naive-drift 1-month forecasts, a per-month cross-series volatility rank, plus the core-vs-headline CPI gap and the core-PCE-vs-headline spread. Covers CPIAUCSL (headline CPI, seasonally adjusted), CPILFESL (core CPI), PCEPILFE (core PCE price index — the Fed's preferred gauge) and PPIACO (producer prices, all commodities). All rows are normalized to country_code USA so they join cleanly with US macro data. Raw series: Federal Reserve Bank of St. Louis (FRED).

    • inflation
    • cpi
    • pce
    • ppi
    lignes
    3 965
    Qualité
    97
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • Japan Macro Signals (derived)

    Japan macro signals (CPI, 10-year JGB yield, deflation gauges)

    Monthly Japan macro signals from OECD Main Economic Indicators (via FRED): headline CPI (JPNCPIALLMINMEI, 1955 ->) and the 10-year JGB yield (IRLTLT01JPM156N, 1989 ->), with 3-month momentum, year-on-year change, 30-month change volatility, 3-sigma anomaly flags, naive-drift forecasts, plus deflation / high-inflation flags, a 10-year-yield 5-year z-score and a negative-yield flag covering the NIRP and yield-curve-control years. The Japan lens on global macro — the deflation laboratory and Asia-Pacific rates anchor. Companion to us-treasury-yield-curve-signals (US) and imf-global-real-gdp-growth-signals (annual, global). All rows normalized to country_code JPN. Raw series: OECD Main Economic Indicators via FRED.

    • japan
    • cpi
    • inflation
    • deflation
    lignes
    1 250
    Qualité
    91
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Licence incertaine

  • JOLTS Labor Market Signals (derived)

    US JOLTS labor market signals (openings, quits, layoffs churn gauge, anomalies)

    Monthly labor-market signals derived from FRED's Job Openings and Labor Turnover Survey (JOLTS): 30-month annualized change volatility, 3-month momentum, 3-sigma anomaly flags vs a trailing 12-month baseline, naive-drift 1-month forecasts, a per-month cross-series volatility rank, and the quits-minus-layoffs churn spread (the tight-vs-loose labor market gauge). Covers job openings level and rate, quits rate, layoffs & discharges rate and hires rate. All rows are normalized to country_code USA so they join cleanly with US macro data. Raw series: Federal Reserve Bank of St. Louis (FRED); underlying survey: U.S. Bureau of Labor Statistics.

    • labor
    • jolts
    • job-openings
    • quits
    lignes
    1 540
    Qualité
    99
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • CISA KEV exploit-priority intelligence (agent-curated)

    CISA KEV exploit-priority ranking (monthly)

    Monthly exploit-priority ranking of every vulnerability in CISA's Known Exploited Vulnerabilities (KEV) catalog. The full KEV feed (cveID, vendor/product, dateAdded, dueDate, ransomware-campaign flag, required action) is joined with the FIRST EPSS daily exploit-probability CSV on CVE id, deduplicated and validated on cveID, with vendor names whitespace-normalized through a deterministic alias map (raw value kept for audit). Each row carries days since added, days to due (negative = overdue), an EPSS tier, and a documented 0-100 priority_score = 100*(0.40*EPSS percentile + 0.25*due-date urgency + 0.20*recency + 0.15*ransomware flag), ranked as exploit_rank. Month-granular as-of stamping makes same-month re-runs hash-identical; the EPSS score_date is recorded per row so the daily vintage is auditable. Columns: snapshot month, fetch timestamp, CVE id, canonical + raw vendor, product, vulnerability name, description, CWE ids, dates, day counts, overdue/ransomware flags, forensic triage, required action, notes, KEV catalog version + provenance URLs, EPSS score/percentile/date/tier, priority score/tier/rank. Primary key: (snapshot_month, cve_id). Cadence: monthly; each snapshot is the full KEV catalog joined with that day's EPSS scores, ranked for remediation triage. Nullability: epss/epss_percentile/epss_tier are null when a KEV CVE is absent from the EPSS CSV (epss_missing=1; contributes 0 to the score — rare, only very fresh catalog additions); nothing else is null. Caveats: priority_score is a triage heuristic, not a risk quantification; EPSS estimates 30-day exploitation probability, not observed exploitation; KEV vendor names are CISA-curated. KEV catalog is US federal government work (public domain); EPSS scores are published free of charge by the FIRST EPSS SIG with no registration — no explicit redistribution license is stated, so commercial_use = unclear with per-source attribution. Sample use: order by exploit_rank for the vulnerabilities to patch first, or filter priority_tier = 'p1'.

    • technology
    • cybersecurity
    • signals
    • us
    lignes
    1 723
    Qualité
    93
    Mis à jour
    25 sept. 2026
    À jour
    Licence
    Licence incertaine

  • Labor-Demand Signals (derived)

    US labor-demand signals (JOLTS openings, hires, quits)

    Monthly US labor-demand signals from BLS JOLTS data (via FRED, 2000 ->): job openings, the hires rate and the quits rate, with 3-month momentum, 30-month change volatility, 3-sigma anomaly flags, drift forecasts, 3-year demand z-scores, a hires hiring-freeze flag, a quits-based worker-confidence gauge and a vacancy-surge flag. The labor-DEMAND companion to the labor-supply levels in us-labor-market-signals: how tight the labor market is from the employer's and the worker's side. All rows normalized to country_code USA. Raw series: Bureau of Labor Statistics (JOLTS) via FRED.

    • labor
    • labor-market
    • jolts
    • job-openings
    lignes
    924
    Qualité
    94
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • Labor Market Signals (derived)

    US labor market signals (Sahm-rule recession indicator, claims anomalies, momentum, forecasts)

    Daily-row labor-market signals derived from FRED's US labor series: the Sahm-rule recession indicator on the unemployment rate (trigger flag + consecutive-month streak), 30-period annualized volatility of changes, 3-month momentum, 3-sigma anomaly flags, naive-drift 1-month forecasts, a per-date cross-series volatility rank, and the 4-week moving average on weekly jobless claims. Covers UNRATE (monthly unemployment rate), PAYEMS (monthly nonfarm payrolls), ICSA (weekly initial claims, seasonally adjusted) and CCSA (weekly continued claims, seasonally adjusted). All rows are normalized to country_code USA so they join cleanly with US macro data. Raw series: Federal Reserve Bank of St. Louis (FRED).

    • labor
    • unemployment
    • jobless-claims
    • sahm-rule
    lignes
    8 224
    Qualité
    97
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • US Labor-Market Slack Signals (derived)

    US labor-market slack signals (duration & underemployment)

    Monthly US labor-market slack signals from BLS duration and underemployment series (FRED UEMPMEAN 1948 ->, U6RATE 1994 ->): 3-month momentum, year-on-year change, 30-month change volatility, 3-sigma anomaly flags, naive-drift forecasts, a long-duration (>=27 weeks) flag, an elevated-underemployment (U-6 >= 10%) flag, and 5-year z-scores for both. The slack lens on US labor — complementing us-labor-market-signals (headline levels) and sahm-labor-recession-signals (recession trigger) by watching unemployment duration and underemployment, where labor-market pain concentrates. All rows normalized to country_code USA. Raw series: U.S. Bureau of Labor Statistics (Current Population Survey) via FRED.

    • unemployment
    • underemployment
    • labor-market
    • duration
    lignes
    1 334
    Qualité
    96
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • Global Living-Standards Signals (derived)

    Global living-standards signals (GDP per capita PPP, ranks, catch-up)

    Yearly global living-standards signals from IMF DataMapper GDP per capita at purchasing power parity (PPPPC, ~190 economies, 1980 ->, WEO projections as published): 5- and 10-year changes, 10-year OLS trend slopes, 3-sigma anomaly flags, linear 5-year forecasts, 10-year income z-scores, per-year global income ranks, high/low-income flags and a catch-up (convergence) flag. The living-standards lens on the world economy — income levels, not growth rates. Companion to imf-global-real-gdp-growth-signals (rates) and global-debt-signals (fiscal stock). All rows carry canonical country_code labels. Raw indicator: International Monetary Fund, DataMapper.

    • living-standards
    • income
    • gdp-per-capita
    • ppp
    lignes
    9 268
    Qualité
    91
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Licence incertaine

  • LLM benchmark scoreboard (agent-curated)

    LLM composite capability index (monthly)

    Monthly per-model composite of the cross-leaderboard scoreboard: for every model appearing in LMArena's official public leaderboard (text/overall, CC-BY-4.0) or Epoch AI's Capabilities & Benchmarking hub (CC-BY), the composite index is the equal-weighted mean of its normalized 0-100 score components (LMArena Bradley-Terry rating rescaled + each Epoch benchmark rescaled), with n_components recorded so single-signal models are distinguishable from broadly measured ones. Epoch's own Capabilities Index (ECI) is carried as a reference column, never a composite input. Columns: snapshot date, composite rank, resolved model key, display name, resolution tier, organization, ISO alpha-3 country, accessibility, component count, Epoch benchmarks covered, LMArena variants merged, Epoch mean normalized, LMArena normalized, LMArena rating/rank/votes, Epoch ECI + CI, composite index, source URLs. Primary key: (snapshot_date, model_key). Cadence: monthly. Caveats: a transparency-first average, not a capability claim; benchmarks differ in difficulty and LMArena measures human preference. Sample use: order by composite_rank for the current cross-source model ranking, or filter n_components >= 5 for broadly-measured models only.

    • ai-research
    • machine-learning
    • technology
    • signals
    lignes
    546
    Qualité
    96
    Mis à jour
    25 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • LLM benchmark scoreboard (agent-curated)

    LLM benchmark scores (monthly)

    Monthly cross-leaderboard scoreboard of large language models: every (model, benchmark) score from LMArena's official public leaderboard dataset (Bradley-Terry ratings, text/overall arena, CC-BY-4.0) and Epoch AI's Capabilities & Benchmarking data hub (56 current capability benchmarks, CC-BY), rescaled per benchmark to a comparable 0-100 (min-max within the snapshot; all benchmarks are higher-is-better) with per-benchmark ranks. Models are entity-resolved across the two sources (exact name match, then one arena-variant suffix stripped, unambiguous only; tier recorded per row). Superseded Epoch benchmarks are pruned. Columns: snapshot date (day-granular UTC), source, benchmark, benchmark release date, resolved model key, raw upstream model id, display name, resolution tier, organization, ISO alpha-3 country, raw score, score unit, normalized 0-100 score, rank in benchmark, models in benchmark, vote count (LMArena), source publish date, source URL. Primary key: (snapshot_date, source, benchmark, source_model_id). Cadence: monthly; same-day re-runs are content-hash no-ops. Caveats: the 0-100 scale is within-benchmark relative, not absolute; LMArena measures human preference, Epoch benchmarks measure task accuracy — the composite dataset blends them explicitly. Sample use: filter benchmark = 'GPQA diamond' order by rank_in_benchmark for the current reasoning-benchmark ranking.

    • ai-research
    • machine-learning
    • technology
    • signals
    lignes
    3 121
    Qualité
    99
    Mis à jour
    25 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • US Loan-Loss Signals (derived)

    US loan-loss signals (charge-off rates, stress regimes)

    Quarterly US bank loan-loss signals from Federal Reserve Board charge-off rates (via FRED): all-real-estate-loan charge-offs (CORALACBS) and credit-card charge-offs (CORCCACBS), 1991 ->, with quarter-on-quarter and year-on-year change, 30-quarter change volatility, 3-sigma anomaly flags, naive-drift forecasts, 5-year loss z-scores, and elevated / severe loss-regime flags. The realized-losses lens on bank credit quality — the realized damage to delinquency's early warning. Companion to us-bank-credit-cycle-signals (volumes, delinquency) and us-sloos-bank-lending-standards-signals (standards). All rows normalized to country_code USA. Raw series: Board of Governors of the Federal Reserve System via FRED.

    • banks
    • charge-offs
    • loan-losses
    • credit-quality
    lignes
    332
    Qualité
    99
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Usage commercial OK

  • Global Longevity Convergence Signals (derived)

    Global longevity convergence signals (frontier gaps, catch-up velocity)

    Value-added longevity signals from the free WHO Global Health Observatory: life-expectancy frontier gaps and catch-up velocity, healthy-life-expectancy trends, suicide-rate trends and malaria decline, with 3-sigma anomaly flags and cross-country ranks across ~190 countries. All computation is local pandas/numpy; no paid models or APIs.

    • longevity
    • life-expectancy
    • mortality
    • who
    lignes
    14 890
    Qualité
    93
    Mis à jour
    22 sept. 2026
    À jour
    Licence
    Usage commercial interdit

Utiliser votre propre clé d’IA

Une fois l’allocation gratuite du jour épuisée, les fonctions d’IA peuvent passer par votre propre compte fournisseur.

Conservée uniquement dans cet onglet (effacée à sa fermeture) et envoyée avec chaque requête d’IA. Nos serveurs l’utilisent pour cette requête et ne la stockent ni ne la journalisent jamais.