spb/wp14_uqo

1% ============================================================================2\section{Data}3\label{sec:data}4% ============================================================================56\subsection{Data Sources}7\label{subsec:data_sources}89We assemble a cross-section of residential listings from two complementary Quebec10platforms, both retrieved on 3~July~2026 via web-scraping actors. The first,11\textbf{Realtor.ca}, is the national Multiple Listing Service (MLS) portal and12provides broker-listed inventory (27{,}302 raw records). The second,13\textbf{DuProprio}, is the dominant for-sale-by-owner platform in Quebec (7{,}69314raw records). Pooling the two sources broadens coverage across price segments and15sale channels---brokered and FSBO listings differ systematically in seller16composition and marketing \citep{hendel2009mls}---and a source indicator absorbs any17systematic level difference between them. The price we observe is the \emph{asking}18price; Section~\ref{subsec:limits} discusses the interpretation of hedonic19coefficients estimated on list rather than transaction prices20\citep{horowitz1992list,knight2002listing}.2122Each record is a structured JSON object containing a price, an address, a23property-attribute block, a free-text description, a list of features, and an array24of image URLs. The two sources differ in their auxiliary fields: Realtor.ca records25include room-by-room dimensions but no geographic coordinates, whereas DuProprio26records include coordinates and captioned images but no room block. We construct a27unique key by concatenating the source and the listing identifier. Because the data28form a single-date cross-section, no time dimension is modeled.2930\subsection{Sample Construction}31\label{subsec:sample}3233We restrict attention to \emph{houses}, defined as single-family dwellings and34townhouses. On Realtor.ca this corresponds to building types \textit{House} and35\textit{Row/Townhouse}; on DuProprio it corresponds to the property types36\textit{Home}, \textit{Bungalow}, \textit{Semi Detached}, \textit{Intergenerational},37\textit{Country Home}, \textit{Cottage}, \textit{Villa}, and \textit{Townhouse}.38Condominiums, income properties (duplex through quintuplex), vacant land, farms,39commercial properties, and mobile homes are excluded. After requiring a non-missing,40positive price, the pooled sample contains 19{,}273 houses.4142\subsection{Outlier Cleaning}43\label{subsec:cleaning}4445Scraped prices contain parsing artifacts spanning several orders of magnitude (the46raw minimum and maximum are \$899 and \$368~billion). We apply conservative,47transparent bounds rather than model-based trimming: prices are retained in the48\$50{,}000--\$15{,}000{,}000 range; living area in 300--20{,}000~ft$^2$; lot area in49200--5{,}000{,}000~ft$^2$; bedrooms in 0--15 and bathrooms in 0--12. When a single50attribute violates its bound it is set to missing (and later imputed) rather than51deleting the observation, preserving sample size. Finally, we trim the52price-per-square-foot distribution at its 0.5th and 99.5th percentiles53(approximately \$73--\$1{,}276/ft$^2$) among listings with a measured area, removing54residual price- or area-parsing errors. The cleaned sample contains55\textbf{19{,}123 houses}, of which 80.3\% originate from Realtor.ca.5657\subsection{Summary Statistics}58\label{subsec:summary_stats}5960Table~\ref{tab:summary_stats} reports descriptive statistics for the key variables.61The median list price is \$599{,}900. The distribution of living area is right-skewed,62motivating the logarithmic transformation adopted in Section~\ref{sec:methodology}.63Living area is reported for only a minority of Realtor.ca listings; where it is64missing we reconstruct it from room-level dimensions (Section~\ref{subsec:structural}),65and where neither is available we rely on the area-free specification.6667\input{tables/summary_stats}6869Table~\ref{tab:ai_summary} previews the machine-generated attributes described in70Section~\ref{sec:methodology}: roughly a quarter of listings are described as recently71renovated, the mean visual condition and curb-appeal scores are 7.4 and 7.2 on a7210-point scale, and 43\% of façades display a visual luxury cue.7374\input{tables/ai_summary}75