Euro area inflation density forecasts
Quantile regression forest. Direct forecasts of NSA year-on-year HICP and HICPex. TreeSHAP bars are deviations from the forest baseline E[f].
Pending predictive quantiles
Forecasts whose outcome is not yet observed. Latest vintage is highlighted.
Contribution to the forecast
Stacked contributions to the forecast minus E[f]. One bar per target month.
Nonlinear inflation scenario simulator
Set future path assumptions. A separate conditional QRF—trained on historically realised future paths—translates them into a density shift while the forecasting QRF remains the baseline. Brackets show the latest observation used by the model.
Checking scenario API…
Baseline and scenario predictive densities will appear here.
Updated feature contributions
TreeSHAP contributions to the forest mean; lags of each series are summed.
ECB policy indicator
Forward-looking Taylor rule i* = r* + π + 0.5(π − π*) + 0.5 y with π* = 2%. QRF rows map the full predictive distribution through the same rule. Output gap is the latest AMECO euro-area estimate.
TreeSHAP groups over the OOS window
Dated by target month. Origins from 2019 onward, as in the paper’s attribution exercise.