Python 52.4%
TeX 47.6%
1% Auto-generated (revision: recomputed on identical folds)2\begin{table}[htbp]3 \centering4 \begin{threeparttable}5 \caption{Feature Ablation: Out-of-Sample $R^2$ by Feature Layer}6 \label{tab:ablation}7 \begin{tabular}{@{}lcccc@{}}8 \toprule9 Feature set & \# feats & OLS & GBM & Blend \\10 \midrule11 Structural + amenity & 24 & 0.521 & 0.692 & --- \\12 \quad + Rooms + listing & 38 & 0.619 & 0.729 & --- \\13 \quad + Text (LLM) & 56 & 0.628 & 0.743 & 0.760 \\14 \quad + Vision (façade) & 77 & 0.686 & 0.770 & 0.779 \\15 \bottomrule16 \end{tabular}17 \begin{tablenotes}[flushleft]18 \footnotesize19 \item \textit{Notes:} Each row adds a feature layer to the row above (nested).20 Entries are out-of-sample $R^2$ on $\ln(\text{Price})$ from 5-fold cross-validation21 with a common seed. ``\# feats'' counts numeric/indicator features before the city,22 type, and source fixed effects (added in every specification). The blend averages23 the robust neural network and gradient boosting and is reported where both were24 trained. GBM handles missing living area natively; OLS uses within-type median imputation.25 The gain from a layer is the difference between consecutive rows; the room and listing26 block includes the Realtor.ca room detail and therefore also carries platform identity.27 \end{tablenotes}28 \end{threeparttable}29\end{table}30