090°Données ouvertes
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.
1 jeux de données
Federal Reserve Bank of New York
US state household-debt stress intelligence from the NY Fed Consumer Credit Panel / Equifax state-level workbook (keyless, redistributable under the NY Fed Terms of Use): 52 geographies x Q4 year, 2003-2025. Core measures: per-capita debt balances for auto, credit card, mortgage, and student loans, plus the percent of each balance 90+ days delinquent. Method: the ten data sheets are parsed from their fixed grid layout and pivoted to a long state x year panel; rows with a NULL total balance are dropped as unverifiable (never fabricated); the allUS national row is integrity-checked then excluded. Derived signals per state-year: a documented 0-100 stress_score (mean of full-panel min-max normalized delinquency rates, equal weights), within-year stress rank and quintile tier, debt burden rank, YoY percent change of total debt per capita, YoY change in credit-card 90+ delinquency (pp), and a severe-stress flag (score >= 75). Use as prediction features: join to customers/orders/leads by region_code + date for demand, churn, and lead-scoring models — household balance-sheet stress leads changes in discretionary spending, subscription churn, and B2B close rates. Caveats: Q4-only annual grain (forward-fill within the year for higher-frequency models); balances are nominal, not inflation-adjusted; delinquency is a stock measure (use the YoY flow signals for deterioration); Puerto Rico is published only through 2016; student-loan sheets use a thinner 1% panel.
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.