CHRIONML® AI LABS / INTELLIGENCE TOUR / WEEK 01

Boston vs Providence | Forecast Lab

Real uploaded monthly condo-price observations · browser-trained ridge-regression model · transparent holdout evaluation and interactive forecasts.

Market forecast

Model trained locally
Last observed price—
Forecast endpoint—
Holdout MAE—
Holdout MAPE—
ObservedRidge forecastOther market (historical)Dashed gray: naive baseline when enabled

Model evaluation and forecast table

MonthForecastNaive baseline

Data provenance & scientific caveats

Source: user-provided Boston and Providence monthly datasets; 80 monthly observations per market, January 2020–August 2026. Prices are reproduced directly from the supplied files. Original upstream source and geographic/condo-series definitions should be independently documented before publication.

Forecasts are generated in your browser from an actual fitted regression model, not hardcoded predictions. Evaluation uses the final 12 historical observations as an untouched test set; model refits on all available data for future forecasts. The forecast is experimental and not financial advice. This version does not estimate prediction intervals or establish superior accuracy against the naive benchmark. The fixed-origin holdout differs from the 44 expanding-window one-month-ahead backtests in the published research; do not compare their error metrics directly.