SPB Git forge
spboucher.ai ↗

spb/wp22_uqo

Public
3commits 1branches 0releases
133.8 MBsize
maindefault branch
1 h agolast push
Python 58.4% TeX 40.8% SQL 0.6%

WP22: benchmarks complete, region/province comparison, revision pipeline with two modes

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Simon-Pierre Boucher committed 1 h ago (Oct 21, 2026) parent 8f2d533

46 changed files +905 −54

modified Makefile +1 −1
@@ -8,7 +8,7 @@ benchmarks:
8 8 hmts:
9 9 $(PY) scripts/11_hmts.py
10 10 simulation:
11 − $(PY) scripts/20_simulation.py --reps 80 --workers 12
11 + $(PY) scripts/20_simulation.py --reps 80 --workers 12; $(PY) scripts/21_mc_summary.py
12 12 panel:
13 13 $(PY) scripts/30_panel.py
14 14 vintages:
added data/processed/index_vintages_filtered_only.parquet +0 −0

Binary file not shown.

modified figures/fig_aggregation.pdf +0 −0

Binary file not shown.

modified figures/fig_arrival_profile.pdf +0 −0

Binary file not shown.

modified figures/fig_composition_shock.pdf +0 −0

Binary file not shown.

modified figures/fig_coverage_monthly.pdf +0 −0

Binary file not shown.

modified figures/fig_gain_vs_n.pdf +0 −0

Binary file not shown.

modified figures/fig_hedonic_profiles.pdf +0 −0

Binary file not shown.

modified figures/fig_map_municipal_house.pdf +0 −0

Binary file not shown.

modified figures/fig_moran.pdf +0 −0

Binary file not shown.

modified figures/fig_municipal_examples.pdf +0 −0

Binary file not shown.

modified figures/fig_municipal_examples.png +0 −0

Binary file not shown.

modified figures/fig_placebo.pdf +0 −0

Binary file not shown.

modified figures/fig_province_methods.pdf +0 −0

Binary file not shown.

modified figures/fig_province_methods.png +0 −0

Binary file not shown.

modified figures/fig_region_methods.pdf +0 −0

Binary file not shown.

modified figures/fig_region_methods.png +0 −0

Binary file not shown.

modified paper/figures/fig_aggregation.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_arrival_profile.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_composition_shock.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_coverage_monthly.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_gain_vs_n.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_hedonic_profiles.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_map_municipal_house.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_moran.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_municipal_examples.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_placebo.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_province_methods.pdf +0 −0

Binary file not shown.

modified paper/figures/fig_region_methods.pdf +0 −0

Binary file not shown.

added paper/main_anon.tex +153 −0
@@ -0,0 +1,153 @@
1 +% ============================================================================
2 +% UQO Working Paper Template — Economics / Finance
3 +% ============================================================================
4 +\documentclass[12pt,letterpaper]{article}
5 +
6 +% --- Encoding & Language ---
7 +\usepackage[utf8]{inputenc}
8 +\usepackage[T1]{fontenc}
9 +\usepackage[english]{babel}
10 +
11 +% --- Page Layout ---
12 +\usepackage[
13 + letterpaper,
14 + top=1in,
15 + bottom=1in,
16 + left=1in,
17 + right=1in
18 +]{geometry}
19 +
20 +% --- Typography ---
21 +\usepackage{mathptmx} % Times-like font (text + math)
22 +\usepackage{microtype} % Improved typographic refinements
23 +\usepackage{setspace}
24 +\onehalfspacing % 1.5 line spacing
25 +
26 +% --- Math ---
27 +\usepackage{amsmath,amssymb,amsthm}
28 +
29 +% --- Tables ---
30 +\usepackage{booktabs}
31 +\usepackage{tabularx}
32 +\usepackage{multirow}
33 +\usepackage{threeparttable}
34 +\usepackage{siunitx}
35 +\sisetup{
36 + input-symbols = {()},
37 + table-format = -1.3,
38 + parse-numbers = false
39 +}
40 +
41 +% --- Figures ---
42 +\usepackage{graphicx}
43 +\usepackage{subcaption}
44 +\usepackage{float}
45 +
46 +% --- Colors & Links ---
47 +\usepackage[dvipsnames]{xcolor}
48 +\usepackage[
49 + colorlinks = true,
50 + linkcolor = NavyBlue,
51 + citecolor = NavyBlue,
52 + urlcolor = NavyBlue,
53 + breaklinks = true
54 +]{hyperref}
55 +\hypersetup{pdftitle={SHARP-HPI: A Spatially Hierarchical, Adaptive and Robust Framework for Quality-Adjusted Housing Price Indices}, pdfauthor={}}
56 +
57 +% --- Bibliography ---
58 +\usepackage[authoryear,round,semicolon]{natbib}
59 +\bibliographystyle{aer}
60 +
61 +% --- Captions ---
62 +\usepackage[
63 + font = small,
64 + labelfont = bf,
65 + labelsep = period,
66 + justification = justified,
67 + singlelinecheck = false
68 +]{caption}
69 +
70 +% --- Headers & Footers ---
71 +\usepackage{fancyhdr}
72 +\pagestyle{fancy}
73 +\fancyhf{}
74 +\renewcommand{\headrulewidth}{0pt}
75 +\fancyfoot[C]{\thepage}
76 +
77 +% --- Section Formatting ---
78 +\usepackage{titlesec}
79 +\titleformat{\section}{\large\bfseries}{\thesection.}{0.5em}{}
80 +\titleformat{\subsection}{\normalsize\bfseries}{\thesubsection.}{0.5em}{}
81 +\titleformat{\subsubsection}{\normalsize\itshape}{\thesubsubsection.}{0.5em}{}
82 +
83 +% --- Appendix Support ---
84 +\usepackage[toc,page]{appendix}
85 +\usepackage{longtable}
86 +\usepackage{bm}
87 +\input{tables/numbers}
88 +\input{tables/numbers_missing}
89 +\newtheorem{proposition}{Proposition}
90 +\newtheorem{assumption}{Assumption}
91 +\newtheorem{definition}{Definition}
92 +\newcommand{\E}{\mathbb{E}}
93 +\newcommand{\Var}{\mathrm{Var}}
94 +\DeclareMathOperator*{\argmax}{arg\,max}
95 +
96 +% --- Misc ---
97 +\usepackage{enumitem}
98 +\setlist{nosep,leftmargin=*}
99 +
100 +% ============================================================================
101 +% METADATA — Edit these fields for each working paper
102 +% ============================================================================
103 +\newcommand{\WPnumber}{22}
104 +\newcommand{\WPtitle}{SHARP-HPI: A Spatially Hierarchical, Adaptive and Robust Framework for Quality-Adjusted Housing Price Indices}
105 +\newcommand{\WPsubtitle}{Monthly municipal indices for Quebec from 770{,}000 registered sales, 2021--2026}
106 +\newcommand{\WPdate}{October 2026}
107 +\newcommand{\WPversion}{1.0}
108 +\input{sections/abstract}
109 +\newcommand{\WPkeywords}{house price index, hedonic regression, state-space model, small-area estimation, empirical Bayes, real-time data, Quebec}
110 +\newcommand{\WPjel}{R31, C43, C11, C21, C55}
111 +
112 +% --- Author ---
113 +\newcommand{\WPauthor}{Simon-Pierre Boucher}
114 +\newcommand{\WPaffiliation}{%
115 + Département des sciences administratives\\
116 + Université du Québec en Outaouais%
117 +}
118 +\newcommand{\WPemail}{simon-pierre.boucher@uqo.ca}
119 +\newcommand{\WPaddress}{%
120 + Gatineau -- Pavillon Alexandre-Taché\\
121 + 283, boulevard Alexandre-Taché\\
122 + Gatineau, Québec, Canada J9A 1L8%
123 +}
124 +
125 +% ============================================================================
126 +% DOCUMENT
127 +% ============================================================================
128 +\begin{document}
129 +
130 +% --- Title Page ---
131 +\input{sections/titlepage_anon}
132 +
133 +% --- Main Body ---
134 +\input{sections/introduction}
135 +\input{sections/literature}
136 +\input{sections/data}
137 +\input{sections/estimand}
138 +\input{sections/framework}
139 +\input{sections/estimation}
140 +\input{sections/simulation}
141 +\input{sections/empirical}
142 +\input{sections/validation}
143 +\input{sections/implications}
144 +\input{sections/conclusion}
145 +
146 +% --- References ---
147 +\newpage
148 +\bibliography{references}
149 +
150 +% --- Appendix ---
151 +\input{appendix/appendix}
152 +
153 +\end{document}
modified paper/sections/validation.tex +17 −1
@@ -2,7 +2,23 @@
2 2
3 3 \subsection{Benchmarks at the provincial and regional levels}\label{sec:bench}
4 4
5 −%%BENCH%%
5 +Table~\ref{tab:benchmarks_province} reports every benchmark at the provincial level: the cumulative log change from January 2021 to September 2026 and the standard deviation of monthly log changes, with wall-clock times. Figure~\ref{fig:provmethods} plots the series. Four facts stand out. First, the quality-adjusted methods agree on the magnitude of the house appreciation: pooled time dummies, rolling time dummies with any window and splice, hedonic imputation, GEKS, HMTS, the QHPI architecture and SHARP-HPI all lie between \ProvDrtdwOneThreemovementhouseTotal{} and \ProvGhmtswOneThreehouseTotal{} log points, against \ProvAmedianhouseTotal{} for the median and \ProvHrepeatsalescshouseTotal{} for Case--Shiller repeat sales. The median understates because the sold houses became smaller and older over the sample; repeat sales overstate because resold houses are renovated. Second, the chained imputation indices drift relative to their direct counterparts for condominiums and plexes (Laspeyres \ProvEhilaspeyreschainedcondoTotal{} against Paasche \ProvEhipaaschechainedcondoTotal{} for condominiums), the classical chain-drift of monthly chaining with per-period hedonic functions estimated on a few thousand sales. Third, methods differ far more in month-to-month volatility than in trend: the standard deviation of monthly changes is about 1.9 log points for the time-dummy family and SHARP, 2.8 for imputation and GEKS, 3--4 for the median and repeat sales, and 1.3--1.9 for HMTS, whose state-space smoothing of each base-period series is visible. Fourth, the gradient-boosting residual index, which forecasts each month's prices from models trained on earlier months, rises by only \ProvJgbmootresidualhouseTotal{} log points: a predictive model that learns location and time jointly absorbs part of the appreciation into its fitted surface, which is why predictive accuracy is not index accuracy. SHARP-HPI with and without time-varying quality prices differ by less than one log point in total at the provincial level; the drift matters below the province.
6 +
7 +\input{tables/tab_benchmarks_province}
8 +
9 +\begin{figure}[htbp]\centering
10 +\includegraphics[width=\textwidth]{figures/fig_province_methods.pdf}
11 +\caption{Provincial indices by method (2021 average $=100$), January 2021 to September 2026.}\label{fig:provmethods}
12 +\end{figure}
13 +
14 +Table~\ref{tab:benchmarks_region} moves to the 16 administrative regions, where sample sizes range from 60 to 1{,}200 sales a month and the methods that estimate a separate hedonic function per region become unstable. The regional comparison has no truth; we report the volatility of each series and its root mean squared distance to the regional Case--Shiller index over 12-month changes. Three results are relevant to the design. The median and stratified median are two to three times as volatile as the hedonic methods and the farthest from repeat sales. Among hedonic methods, the QHPI architecture, which fits each liquid region with its own coefficients and smooths lightly, is the least volatile (\RegSdKqhpidirectlocaltdhouse{} for houses) and the closest to repeat sales (\RegRmseRSKqhpidirectlocaltdhouse); SHARP-HPI, whose regional path is the pooled provincial path plus a smoothed deviation, is less volatile than the pooled time dummy (\RegSdSsharphouse{} against \RegSdCpooledtdhouse) but not closer to repeat sales than it (\RegRmseRSSsharphouse{} against \RegRmseRSCpooledtdhouse). The reason is identifiable: with 400 to 1{,}200 sales a month the regional Kalman gain is close to one, so the SHARP regional path is essentially the pooled model's monthly regional mean, and the pooled model constrains quality prices to be common across regions. The Monte Carlo experiments of Section~\ref{sec:simulation}, where the truth is known, show that this costs little when quality prices are common and that regional estimates are then close to the truth; on the real data the repeat-sales arbiter, biased though it is, suggests that region-specific quality prices carry information at the regional level. Figure~\ref{fig:regmethods} shows six regions.
15 +
16 +\input{tables/tab_benchmarks_region}
17 +
18 +\begin{figure}[htbp]\centering
19 +\includegraphics[width=\textwidth]{figures/fig_region_methods.pdf}
20 +\caption{Regional house indices by method for six regions of decreasing liquidity (2021 $=100$).}\label{fig:regmethods}
21 +\end{figure}
6 22
7 23 \subsection{Agreement with repeat sales}
8 24
modified paper/tables/numbers.tex +131 −0
@@ -248,6 +248,15 @@
248 248 \newcommand{\PlaceboFPplexmrc}{0.2}
249 249 \newcommand{\PlaceboFPplexmunicipality}{0.1}
250 250 \newcommand{\PlaceboFPplexregion}{0.3}
251 +\newcommand{\PolFirstcondo}{0.16}
252 +\newcommand{\PolFirsthouse}{0.16}
253 +\newcommand{\PolFirstplex}{0.26}
254 +\newcommand{\PolMeanAbscondo}{0.77}
255 +\newcommand{\PolMeanAbshouse}{1.50}
256 +\newcommand{\PolMeanAbsplex}{1.53}
257 +\newcommand{\PolThirdcondo}{0.15}
258 +\newcommand{\PolThirdhouse}{0.16}
259 +\newcommand{\PolThirdplex}{0.26}
251 260 \newcommand{\PriorTaucondomrc}{1.89}
252 261 \newcommand{\PriorTaucondomunicipality}{1.18}
253 262 \newcommand{\PriorTaucondoregion}{1.89}
@@ -266,32 +275,72 @@
266 275 \newcommand{\PriorVplexmrc}{7.6}
267 276 \newcommand{\PriorVplexmunicipality}{3.7}
268 277 \newcommand{\PriorVplexregion}{0.1}
278 +\newcommand{\ProvAmeancondoTotal}{36.3}
269 279 \newcommand{\ProvAmeanhouseTotal}{47.8}
280 +\newcommand{\ProvAmeanplexTotal}{43.9}
281 +\newcommand{\ProvAmediancondoTotal}{36.0}
270 282 \newcommand{\ProvAmedianhouseTotal}{46.4}
283 +\newcommand{\ProvAmedianplexTotal}{50.9}
284 +\newcommand{\ProvBstratifiedmediancondoTotal}{37.3}
271 285 \newcommand{\ProvBstratifiedmedianhouseTotal}{54.0}
286 +\newcommand{\ProvBstratifiedmedianplexTotal}{49.6}
287 +\newcommand{\ProvCpooledtdcondoTotal}{41.0}
272 288 \newcommand{\ProvCpooledtdhouseTotal}{56.5}
289 +\newcommand{\ProvCpooledtdolscondoTotal}{40.9}
273 290 \newcommand{\ProvCpooledtdolshouseTotal}{58.5}
291 +\newcommand{\ProvCpooledtdolsplexTotal}{48.7}
292 +\newcommand{\ProvCpooledtdplexTotal}{51.4}
293 +\newcommand{\ProvDrtdwOneThreemeancondoTotal}{37.2}
274 294 \newcommand{\ProvDrtdwOneThreemeanhouseTotal}{55.6}
295 +\newcommand{\ProvDrtdwOneThreemeanplexTotal}{50.6}
296 +\newcommand{\ProvDrtdwOneThreemovementcondoTotal}{37.1}
275 297 \newcommand{\ProvDrtdwOneThreemovementhouseTotal}{55.0}
298 +\newcommand{\ProvDrtdwOneThreemovementplexTotal}{51.8}
299 +\newcommand{\ProvDrtdwOneThreewindowcondoTotal}{38.7}
276 300 \newcommand{\ProvDrtdwOneThreewindowhouseTotal}{56.5}
301 +\newcommand{\ProvDrtdwOneThreewindowplexTotal}{52.0}
302 +\newcommand{\ProvDrtdwSevenmeancondoTotal}{37.3}
277 303 \newcommand{\ProvDrtdwSevenmeanhouseTotal}{56.1}
304 +\newcommand{\ProvDrtdwSevenmeanplexTotal}{51.0}
305 +\newcommand{\ProvDrtdwTwoFivemeancondoTotal}{39.4}
278 306 \newcommand{\ProvDrtdwTwoFivemeanhouseTotal}{56.4}
307 +\newcommand{\ProvDrtdwTwoFivemeanplexTotal}{51.1}
308 +\newcommand{\ProvEhifisherchainedcondoTotal}{41.1}
279 309 \newcommand{\ProvEhifisherchainedhouseTotal}{56.5}
310 +\newcommand{\ProvEhifisherchainedplexTotal}{56.7}
311 +\newcommand{\ProvEhifisherdirectcondoTotal}{41.5}
280 312 \newcommand{\ProvEhifisherdirecthouseTotal}{59.0}
313 +\newcommand{\ProvEhifisherdirectplexTotal}{50.1}
314 +\newcommand{\ProvEhilaspeyreschainedcondoTotal}{45.4}
281 315 \newcommand{\ProvEhilaspeyreschainedhouseTotal}{56.4}
316 +\newcommand{\ProvEhilaspeyreschainedplexTotal}{55.6}
317 +\newcommand{\ProvEhipaaschechainedcondoTotal}{36.8}
282 318 \newcommand{\ProvEhipaaschechainedhouseTotal}{56.5}
319 +\newcommand{\ProvEhipaaschechainedplexTotal}{57.8}
320 +\newcommand{\ProvFgekswOneThreemovementcondoTotal}{41.0}
283 321 \newcommand{\ProvFgekswOneThreemovementhouseTotal}{58.8}
322 +\newcommand{\ProvFgekswOneThreemovementplexTotal}{52.3}
323 +\newcommand{\ProvFgekswTwoFivemovementcondoTotal}{41.1}
284 324 \newcommand{\ProvFgekswTwoFivemovementhouseTotal}{59.0}
325 +\newcommand{\ProvFgekswTwoFivemovementplexTotal}{51.8}
285 326 \newcommand{\ProvGhmtswFourcondoTotal}{42.4}
286 327 \newcommand{\ProvGhmtswFourhouseTotal}{59.2}
287 328 \newcommand{\ProvGhmtswFourplexTotal}{53.2}
288 329 \newcommand{\ProvGhmtswOneThreecondoTotal}{42.2}
289 330 \newcommand{\ProvGhmtswOneThreehouseTotal}{60.1}
290 331 \newcommand{\ProvGhmtswOneThreeplexTotal}{54.9}
332 +\newcommand{\ProvHrepeatsalesbmncondoTotal}{38.6}
291 333 \newcommand{\ProvHrepeatsalesbmnhouseTotal}{68.6}
334 +\newcommand{\ProvHrepeatsalesbmnplexTotal}{57.9}
335 +\newcommand{\ProvHrepeatsalescscondoTotal}{37.6}
292 336 \newcommand{\ProvHrepeatsalescshouseTotal}{67.6}
337 +\newcommand{\ProvHrepeatsalescsplexTotal}{58.5}
338 +\newcommand{\ProvJgbmootresidualcondoTotal}{36.0}
293 339 \newcommand{\ProvJgbmootresidualhouseTotal}{46.9}
340 +\newcommand{\ProvJgbmootresidualplexTotal}{39.3}
341 +\newcommand{\ProvKqhpicondoTotal}{37.2}
294 342 \newcommand{\ProvKqhpihouseTotal}{55.6}
343 +\newcommand{\ProvKqhpiplexTotal}{50.6}
295 344 \newcommand{\ProvPeakMonthcondo}{2026-10}
296 345 \newcommand{\ProvPeakMonthhouse}{2026-06}
297 346 \newcommand{\ProvPeakMonthplex}{2026-06}
@@ -301,8 +350,12 @@
301 350 \newcommand{\ProvSepcondo}{136.4}
302 351 \newcommand{\ProvSephouse}{154.9}
303 352 \newcommand{\ProvSepplex}{151.2}
353 +\newcommand{\ProvSsharpcondoTotal}{41.5}
304 354 \newcommand{\ProvSsharphouseTotal}{56.9}
355 +\newcommand{\ProvSsharpnotvcondoTotal}{41.0}
305 356 \newcommand{\ProvSsharpnotvhouseTotal}{56.7}
357 +\newcommand{\ProvSsharpnotvplexTotal}{51.3}
358 +\newcommand{\ProvSsharpplexTotal}{50.9}
306 359 \newcommand{\ProvYoySepcondo}{1.3}
307 360 \newcommand{\ProvYoySephouse}{6.1}
308 361 \newcommand{\ProvYoySepplex}{4.8}
@@ -326,6 +379,84 @@
326 379 \newcommand{\RSGapTotalregioncondo}{-22.9}
327 380 \newcommand{\RSGapTotalregionhouse}{-0.1}
328 381 \newcommand{\RSGapTotalregionplex}{+3.7}
382 +\newcommand{\RegRmseRSAmediancondo}{13.0}
383 +\newcommand{\RegRmseRSAmedianhouse}{13.7}
384 +\newcommand{\RegRmseRSAmedianplex}{27.5}
385 +\newcommand{\RegRmseRSBstratifiedmediancondo}{12.2}
386 +\newcommand{\RegRmseRSBstratifiedmedianhouse}{13.2}
387 +\newcommand{\RegRmseRSBstratifiedmedianplex}{26.0}
388 +\newcommand{\RegRmseRSCpooledtdcondo}{11.1}
389 +\newcommand{\RegRmseRSCpooledtdhouse}{12.6}
390 +\newcommand{\RegRmseRSCpooledtdplex}{24.6}
391 +\newcommand{\RegRmseRSDrtdwOneThreemeancondo}{12.0}
392 +\newcommand{\RegRmseRSDrtdwOneThreemeanhouse}{12.8}
393 +\newcommand{\RegRmseRSDrtdwOneThreemeanplex}{24.7}
394 +\newcommand{\RegRmseRSEhifisherchainedcondo}{7.6}
395 +\newcommand{\RegRmseRSEhifisherchainedhouse}{12.0}
396 +\newcommand{\RegRmseRSEhifisherchainedplex}{17.8}
397 +\newcommand{\RegRmseRSFgekswTwoFivemovementcondo}{6.5}
398 +\newcommand{\RegRmseRSFgekswTwoFivemovementhouse}{9.7}
399 +\newcommand{\RegRmseRSFgekswTwoFivemovementplex}{13.7}
400 +\newcommand{\RegRmseRSGhmtswOneThreecondo}{6.2}
401 +\newcommand{\RegRmseRSGhmtswOneThreehouse}{10.1}
402 +\newcommand{\RegRmseRSGhmtswOneThreeplex}{14.2}
403 +\newcommand{\RegRmseRSKqhpidirectlocaltdcondo}{6.6}
404 +\newcommand{\RegRmseRSKqhpidirectlocaltdhouse}{9.9}
405 +\newcommand{\RegRmseRSKqhpidirectlocaltdplex}{14.5}
406 +\newcommand{\RegRmseRSSsharpcondo}{11.0}
407 +\newcommand{\RegRmseRSSsharpfilteredcondo}{11.2}
408 +\newcommand{\RegRmseRSSsharpfilteredhouse}{12.8}
409 +\newcommand{\RegRmseRSSsharpfilteredplex}{24.7}
410 +\newcommand{\RegRmseRSSsharphouse}{12.7}
411 +\newcommand{\RegRmseRSSsharpplex}{24.8}
412 +\newcommand{\RegSdAmediancondo}{9.17}
413 +\newcommand{\RegSdAmedianhouse}{7.91}
414 +\newcommand{\RegSdAmedianplex}{13.27}
415 +\newcommand{\RegSdBstratifiedmediancondo}{6.21}
416 +\newcommand{\RegSdBstratifiedmedianhouse}{6.16}
417 +\newcommand{\RegSdBstratifiedmedianplex}{10.00}
418 +\newcommand{\RegSdCpooledtdcondo}{4.65}
419 +\newcommand{\RegSdCpooledtdhouse}{4.41}
420 +\newcommand{\RegSdCpooledtdplex}{7.11}
421 +\newcommand{\RegSdDrtdwOneThreemeancondo}{5.11}
422 +\newcommand{\RegSdDrtdwOneThreemeanhouse}{4.87}
423 +\newcommand{\RegSdDrtdwOneThreemeanplex}{8.91}
424 +\newcommand{\RegSdEhifisherchainedcondo}{3.79}
425 +\newcommand{\RegSdEhifisherchainedhouse}{5.42}
426 +\newcommand{\RegSdEhifisherchainedplex}{7.23}
427 +\newcommand{\RegSdFgekswTwoFivemovementcondo}{3.69}
428 +\newcommand{\RegSdFgekswTwoFivemovementhouse}{5.19}
429 +\newcommand{\RegSdFgekswTwoFivemovementplex}{6.30}
430 +\newcommand{\RegSdGhmtswOneThreecondo}{0.95}
431 +\newcommand{\RegSdGhmtswOneThreehouse}{1.59}
432 +\newcommand{\RegSdGhmtswOneThreeplex}{1.09}
433 +\newcommand{\RegSdKqhpidirectlocaltdcondo}{1.48}
434 +\newcommand{\RegSdKqhpidirectlocaltdhouse}{2.33}
435 +\newcommand{\RegSdKqhpidirectlocaltdplex}{2.20}
436 +\newcommand{\RegSdSsharpcondo}{1.60}
437 +\newcommand{\RegSdSsharpfilteredcondo}{2.28}
438 +\newcommand{\RegSdSsharpfilteredhouse}{4.42}
439 +\newcommand{\RegSdSsharpfilteredplex}{3.61}
440 +\newcommand{\RegSdSsharphouse}{3.75}
441 +\newcommand{\RegSdSsharpplex}{2.56}
442 +\newcommand{\RevFirstTwelvecondomunicipality}{3.59}
443 +\newcommand{\RevFirstTwelvecondoprovince}{0.25}
444 +\newcommand{\RevFirstTwelvecondoregion}{1.93}
445 +\newcommand{\RevFirstTwelvehousemunicipality}{2.42}
446 +\newcommand{\RevFirstTwelvehouseprovince}{0.13}
447 +\newcommand{\RevFirstTwelvehouseregion}{2.24}
448 +\newcommand{\RevFirstTwelveplexmunicipality}{5.26}
449 +\newcommand{\RevFirstTwelveplexprovince}{0.35}
450 +\newcommand{\RevFirstTwelveplexregion}{2.05}
451 +\newcommand{\RevFirstcondomunicipality}{1.82}
452 +\newcommand{\RevFirstcondoprovince}{0.16}
453 +\newcommand{\RevFirstcondoregion}{0.98}
454 +\newcommand{\RevFirsthousemunicipality}{2.31}
455 +\newcommand{\RevFirsthouseprovince}{0.16}
456 +\newcommand{\RevFirsthouseregion}{1.93}
457 +\newcommand{\RevFirstplexmunicipality}{3.03}
458 +\newcommand{\RevFirstplexprovince}{0.26}
459 +\newcommand{\RevFirstplexregion}{1.37}
329 460 \newcommand{\RollSpatialAgree}{99.96}
330 461 \newcommand{\RtwoWithincondo}{0.63}
331 462 \newcommand{\RtwoWithinhouse}{0.53}
modified paper/tables/tab_benchmarks_province.tex +22 −22
@@ -6,30 +6,30 @@
6 6 \toprule
7 7 Method & H: total & H: s.d. $\Delta$ & C: total & C: s.d. $\Delta$ & P: total & P: s.d. $\Delta$ & sec. \\
8 8 \midrule
9 −Median price & 46.4 & 3.97 & — & — & — & — & 0 \\
10 −Mean price & 47.8 & 3.79 & — & — & — & — & 0 \\
11 −Stratified median & 54.0 & 3.35 & — & — & — & — & 0 \\
12 −Pooled time dummy (OLS) & 58.5 & 2.58 & — & — & — & — & 1 \\
13 −Pooled time dummy (Huber) & 56.5 & 1.86 & — & — & — & — & 4 \\
14 −Rolling TD, 7 m, mean splice & 56.1 & 1.89 & — & — & — & — & 12 \\
15 −Rolling TD, 13 m, mean splice & 55.6 & 1.87 & — & — & — & — & 18 \\
16 −Rolling TD, 13 m, movement splice & 55.0 & 1.86 & — & — & — & — & 18 \\
17 −Rolling TD, 13 m, window splice & 56.5 & 1.91 & — & — & — & — & 18 \\
18 −Rolling TD, 25 m, mean splice & 56.4 & 1.86 & — & — & — & — & 32 \\
19 −Hedonic imputation Laspeyres (chained) & 56.4 & 2.76 & — & — & — & — & 5 \\
20 −Hedonic imputation Paasche (chained) & 56.5 & 2.76 & — & — & — & — & 5 \\
21 −Hedonic imputation Fisher (chained) & 56.5 & 2.75 & — & — & — & — & 4 \\
22 −Hedonic imputation Fisher (direct) & 59.0 & 2.76 & — & — & — & — & 4 \\
23 −GEKS-Fisher, 13 m & 58.8 & 2.78 & — & — & — & — & 16 \\
24 −GEKS-Fisher, 25 m & 59.0 & 2.78 & — & — & — & — & 27 \\
9 +Median price & 46.4 & 3.97 & 36.0 & 2.09 & 50.9 & 5.74 & 0 \\
10 +Mean price & 47.8 & 3.79 & 36.3 & 2.47 & 43.9 & 4.16 & 0 \\
11 +Stratified median & 54.0 & 3.35 & 37.3 & 1.88 & 49.6 & 4.61 & 0 \\
12 +Pooled time dummy (OLS) & 58.5 & 2.58 & 40.9 & 1.35 & 48.7 & 2.42 & 1 \\
13 +Pooled time dummy (Huber) & 56.5 & 1.86 & 41.0 & 1.06 & 51.4 & 1.94 & 4 \\
14 +Rolling TD, 7 m, mean splice & 56.1 & 1.89 & 37.3 & 1.01 & 51.0 & 2.06 & 12 \\
15 +Rolling TD, 13 m, mean splice & 55.6 & 1.87 & 37.2 & 1.02 & 50.6 & 1.91 & 18 \\
16 +Rolling TD, 13 m, movement splice & 55.0 & 1.86 & 37.1 & 1.00 & 51.8 & 1.91 & 18 \\
17 +Rolling TD, 13 m, window splice & 56.5 & 1.91 & 38.7 & 1.13 & 52.0 & 2.04 & 18 \\
18 +Rolling TD, 25 m, mean splice & 56.4 & 1.86 & 39.4 & 1.03 & 51.1 & 1.87 & 32 \\
19 +Hedonic imputation Laspeyres (chained) & 56.4 & 2.76 & 45.4 & 1.50 & 55.6 & 2.44 & 5 \\
20 +Hedonic imputation Paasche (chained) & 56.5 & 2.76 & 36.8 & 1.45 & 57.8 & 2.53 & 5 \\
21 +Hedonic imputation Fisher (chained) & 56.5 & 2.75 & 41.1 & 1.45 & 56.7 & 2.44 & 4 \\
22 +Hedonic imputation Fisher (direct) & 59.0 & 2.76 & 41.5 & 1.69 & 50.1 & 2.30 & 4 \\
23 +GEKS-Fisher, 13 m & 58.8 & 2.78 & 41.0 & 1.47 & 52.3 & 2.48 & 16 \\
24 +GEKS-Fisher, 25 m & 59.0 & 2.78 & 41.1 & 1.47 & 51.8 & 2.45 & 27 \\
25 25 HMTS, 4 m & 59.2 & 1.93 & 42.4 & 1.06 & 53.2 & 1.56 & — \\
26 26 HMTS, 13 m & 60.1 & 1.92 & 42.2 & 0.90 & 54.9 & 1.29 & — \\
27 −Repeat sales (BMN) & 68.6 & 3.02 & — & — & — & — & 0 \\
28 −Repeat sales (Case–Shiller) & 67.6 & 2.97 & — & — & — & — & 0 \\
29 −GBM out-of-time residual & 46.9 & 2.54 & — & — & — & — & 191 \\
30 −QHPI v2.1 architecture & 55.6 & 1.87 & — & — & — & — & 18 \\
31 −SHARP stage 1 without time-varying prices & 56.7 & 1.91 & — & — & — & — & 21 \\
32 −SHARP-HPI & 56.9 & 1.91 & — & — & — & — & 21 \\
27 +Repeat sales (BMN) & 68.6 & 3.02 & 38.6 & 1.71 & 57.9 & 4.58 & 0 \\
28 +Repeat sales (Case–Shiller) & 67.6 & 2.97 & 37.6 & 1.73 & 58.5 & 4.61 & 0 \\
29 +GBM out-of-time residual & 46.9 & 2.54 & 36.0 & 1.46 & 39.3 & 2.32 & 191 \\
30 +QHPI v2.1 architecture & 55.6 & 1.87 & 37.2 & 1.02 & 50.6 & 1.91 & 18 \\
31 +SHARP stage 1 without time-varying prices & 56.7 & 1.91 & 41.0 & 1.06 & 51.3 & 1.98 & 21 \\
32 +SHARP-HPI & 56.9 & 1.91 & 41.5 & 1.04 & 50.9 & 1.97 & 21 \\
33 33 \bottomrule
34 34 \end{tabular}
35 35 \begin{tablenotes}\footnotesize\item H = houses, C = condominiums, P = plexes. Rolling TD = rolling time dummy; GEKS on bilateral double-imputation Fisher links with movement splice; HMTS = hedonic multilateral time series (Ishaak et al., 2024); QHPI = re-implementation of the QWHPI v2.1 architecture; last column = wall-clock seconds for the house sample on one Apple M-series core.\end{tablenotes}
modified paper/tables/tab_benchmarks_region.tex +10 −10
@@ -6,16 +6,16 @@
6 6 \toprule
7 7 Method & H: s.d. $\Delta$ & H: RMSE vs RS & C: s.d. $\Delta$ & C: RMSE vs RS & P: s.d. $\Delta$ & P: RMSE vs RS \\
8 8 \midrule
9 −Median price & 7.91 & 13.7 & — & — & — & — \\
10 −Stratified median & 6.16 & 13.2 & — & — & — & — \\
11 −Pooled time dummy (Huber) & 4.41 & 12.6 & — & — & — & — \\
12 −Rolling TD, 13 m, mean splice & 4.87 & 12.8 & — & — & — & — \\
13 −Hedonic imputation Fisher (chained) & 5.42 & 12.0 & — & — & — & — \\
14 −GEKS-Fisher, 25 m & 5.19 & 9.7 & — & — & — & — \\
15 −HMTS, 13 m & nan & 10.1 & nan & — & nan & — \\
16 −QHPI v2.1 architecture & 2.33 & 9.9 & — & — & — & — \\
17 −SHARP-HPI (real-time filtered) & 4.42 & 12.8 & — & — & — & — \\
18 −SHARP-HPI & 3.75 & 12.7 & — & — & — & — \\
9 +Median price & 7.91 & 13.7 & 9.17 & 13.0 & 13.27 & 27.5 \\
10 +Stratified median & 6.16 & 13.2 & 6.21 & 12.2 & 10.00 & 26.0 \\
11 +Pooled time dummy (Huber) & 4.41 & 12.6 & 4.65 & 11.1 & 7.11 & 24.6 \\
12 +Rolling TD, 13 m, mean splice & 4.87 & 12.8 & 5.11 & 12.0 & 8.91 & 24.7 \\
13 +Hedonic imputation Fisher (chained) & 5.42 & 12.0 & 3.79 & 7.6 & 7.23 & 17.8 \\
14 +GEKS-Fisher, 25 m & 5.19 & 9.7 & 3.69 & 6.5 & 6.30 & 13.7 \\
15 +HMTS, 13 m & 1.59 & 10.1 & 0.95 & 6.2 & 1.09 & 14.2 \\
16 +QHPI v2.1 architecture & 2.33 & 9.9 & 1.48 & 6.6 & 2.20 & 14.5 \\
17 +SHARP-HPI (real-time filtered) & 4.42 & 12.8 & 2.28 & 11.2 & 3.61 & 24.7 \\
18 +SHARP-HPI & 3.75 & 12.7 & 1.60 & 11.0 & 2.56 & 24.8 \\
19 19 \bottomrule
20 20 \end{tabular}
21 21 \begin{tablenotes}\footnotesize\item Repeat sales are an independent (pair-based) arbiter, not the truth: they carry renovation and selection biases. Lower s.d. is not better per se; the Monte Carlo (Section 7) separates noise from signal.\end{tablenotes}
added paper/tables/tab_policies.tex +16 −0
@@ -0,0 +1,16 @@
1 +\begin{table}[htbp]
2 +\centering\small
3 +\caption{Provincial monthly change (×100): first and third release versus final}\label{tab:policies}
4 +\begin{threeparttable}
5 +\begin{tabular}{lrrrrrrr}
6 +\toprule
7 +Class & Months & MAE first release & Bias first release & MAE third release & RMSE first & RMSE third & Mean $|\Delta|$ \\
8 +\midrule
9 +condo & 52 & 0.16 & +0.013 & 0.15 & 0.21 & 0.20 & 0.77 \\
10 +house & 52 & 0.16 & -0.013 & 0.16 & 0.20 & 0.20 & 1.50 \\
11 +plex & 52 & 0.26 & -0.014 & 0.26 & 0.35 & 0.35 & 1.53 \\
12 +\bottomrule
13 +\end{tabular}
14 +\begin{tablenotes}\footnotesize\item Full re-estimation policy (smoothed at each vintage). A revision-limited policy freezes the value at the third release; the frozen-first-release (movement-splice) policy is in Table~\ref{tab:revisions_frozen}.\end{tablenotes}
15 +\end{threeparttable}
16 +\end{table}
added paper/tables/tab_revisions.tex +22 −0
@@ -0,0 +1,22 @@
1 +\begin{table}[htbp]
2 +\centering\footnotesize
3 +\caption{Real-time vintages: mean absolute revision of the monthly log change (×100) between the $k$-th release and the final retrospective estimate}\label{tab:revisions}
4 +\begin{threeparttable}
5 +\begin{tabular}{lrrrrrrr}
6 +\toprule
7 +Class / level & Rel. 1 & Rel. 2 & Rel. 3 & Rel. 6 & Rel. 12 & 12-m change, rel. 1 & \% $|$rev$|>1$ pt \\
8 +\midrule
9 +house / province & 0.16 & 0.17 & 0.16 & 0.15 & 0.15 & 0.13 & 0 \\
10 +house / region & 1.93 & 1.95 & 1.96 & 1.91 & 1.94 & 2.24 & 53 \\
11 +house / municipality & 2.31 & 2.31 & 2.33 & 2.35 & 2.43 & 2.42 & 65 \\
12 +condo / province & 0.16 & 0.16 & 0.16 & 0.17 & 0.16 & 0.25 & 0 \\
13 +condo / region & 0.98 & 0.98 & 1.00 & 1.04 & 0.98 & 1.93 & 34 \\
14 +condo / municipality & 1.82 & 1.83 & 1.83 & 1.88 & 1.87 & 3.59 & 54 \\
15 +plex / province & 0.26 & 0.25 & 0.26 & 0.27 & 0.24 & 0.35 & 0 \\
16 +plex / region & 1.37 & 1.38 & 1.36 & 1.35 & 1.41 & 2.05 & 53 \\
17 +plex / municipality & 3.03 & 3.03 & 3.03 & 3.05 & 3.00 & 5.26 & 73 \\
18 +\bottomrule
19 +\end{tabular}
20 +\begin{tablenotes}\footnotesize\item Releases simulated monthly from July 2022 to October 2026 with a 10-day publication delay; municipalities with at least 1,500 sales. Final = smoothed full-sample panel.\end{tablenotes}
21 +\end{threeparttable}
22 +\end{table}
modified results/benchmarks/series.parquet +0 −0

Binary file not shown.

modified results/benchmarks/timing.csv +334 −0
@@ -194,3 +194,337 @@ house,region,17,H_repeat_sales_cs,0.06
194 194 house,region,17,K_qhpi_direct_local_td,0.9
195 195 house,region,17,S_sharp,0.0
196 196 house,region,17,S_sharp_filtered,0.0
197 +condo,province,QC,A_median,0.02
198 +condo,province,QC,A_mean,0.01
199 +condo,province,QC,B_stratified_median,0.08
200 +condo,province,QC,C_pooled_td,1.59
201 +condo,province,QC,C_pooled_td_ols,0.42
202 +condo,province,QC,D_rtd_w13_mean,15.03
203 +condo,province,QC,D_rtd_w13_movement,15.19
204 +condo,province,QC,D_rtd_w13_window,15.69
205 +condo,province,QC,D_rtd_w25_mean,19.94
206 +condo,province,QC,D_rtd_w7_mean,13.26
207 +condo,province,QC,E_hi_laspeyres_chained,9.36
208 +condo,province,QC,E_hi_paasche_chained,9.52
209 +condo,province,QC,E_hi_fisher_chained,9.42
210 +condo,province,QC,E_hi_fisher_direct,9.23
211 +condo,province,QC,F_geks_w25_movement,53.6
212 +condo,province,QC,F_geks_w13_movement,32.0
213 +condo,province,QC,H_repeat_sales_cs,0.25
214 +condo,province,QC,H_repeat_sales_bmn,0.22
215 +condo,province,QC,J_gbm_oot_residual,43819.94
216 +condo,province,QC,K_qhpi,18.44
217 +condo,province,QC,S_sharp,11.9
218 +condo,province,QC,S_sharp_notv,11.6
219 +condo,region,01,A_median,0.0
220 +condo,region,01,B_stratified_median,0.01
221 +condo,region,01,C_pooled_td,0.24
222 +condo,region,01,D_rtd_w13_mean,7.75
223 +condo,region,01,E_hi_fisher,0.24
224 +condo,region,01,F_geks_w25,0.22
225 +condo,region,01,H_repeat_sales_cs,0.03
226 +condo,region,01,K_qhpi_hierarchical,0.02
227 +condo,region,01,S_sharp,0.0
228 +condo,region,01,S_sharp_filtered,0.0
229 +condo,region,02,A_median,0.0
230 +condo,region,02,B_stratified_median,0.01
231 +condo,region,02,C_pooled_td,0.09
232 +condo,region,02,D_rtd_w13_mean,4.75
233 +condo,region,02,E_hi_fisher,0.2
234 +condo,region,02,F_geks_w25,0.23
235 +condo,region,02,H_repeat_sales_cs,0.02
236 +condo,region,02,K_qhpi_hierarchical,0.01
237 +condo,region,02,S_sharp,0.0
238 +condo,region,02,S_sharp_filtered,0.0
239 +condo,region,03,A_median,0.01
240 +condo,region,03,B_stratified_median,0.02
241 +condo,region,03,C_pooled_td,0.25
242 +condo,region,03,D_rtd_w13_mean,6.24
243 +condo,region,03,E_hi_fisher_chained,6.07
244 +condo,region,03,F_geks_w25_movement,35.6
245 +condo,region,03,H_repeat_sales_cs,0.04
246 +condo,region,03,K_qhpi_direct_local_td,0.63
247 +condo,region,03,S_sharp,0.0
248 +condo,region,03,S_sharp_filtered,0.0
249 +condo,region,04,A_median,0.01
250 +condo,region,04,B_stratified_median,0.01
251 +condo,region,04,C_pooled_td,0.11
252 +condo,region,04,D_rtd_w13_mean,7.3
253 +condo,region,04,E_hi_fisher_chained,0.45
254 +condo,region,04,F_geks_w25_movement,0.83
255 +condo,region,04,H_repeat_sales_cs,0.05
256 +condo,region,04,K_qhpi_hierarchical,0.03
257 +condo,region,04,S_sharp,0.0
258 +condo,region,04,S_sharp_filtered,0.0
259 +condo,region,05,A_median,0.01
260 +condo,region,05,B_stratified_median,0.06
261 +condo,region,05,C_pooled_td,0.5
262 +condo,region,05,D_rtd_w13_mean,10.38
263 +condo,region,05,E_hi_fisher_chained,4.94
264 +condo,region,05,F_geks_w25_movement,28.4
265 +condo,region,05,H_repeat_sales_cs,0.02
266 +condo,region,05,K_qhpi_direct_local_td,0.46
267 +condo,region,05,S_sharp,0.0
268 +condo,region,05,S_sharp_filtered,0.0
269 +condo,region,06,A_median,0.01
270 +condo,region,06,B_stratified_median,0.02
271 +condo,region,06,C_pooled_td,0.51
272 +condo,region,06,D_rtd_w13_mean,7.13
273 +condo,region,06,E_hi_fisher_chained,6.22
274 +condo,region,06,F_geks_w25_movement,35.03
275 +condo,region,06,H_repeat_sales_cs,0.08
276 +condo,region,06,K_qhpi_direct_local_td,1.05
277 +condo,region,06,S_sharp,0.0
278 +condo,region,06,S_sharp_filtered,0.0
279 +condo,region,07,A_median,0.0
280 +condo,region,07,B_stratified_median,0.01
281 +condo,region,07,C_pooled_td,0.12
282 +condo,region,07,D_rtd_w13_mean,5.06
283 +condo,region,07,E_hi_fisher_chained,4.87
284 +condo,region,07,F_geks_w25_movement,29.16
285 +condo,region,07,H_repeat_sales_cs,0.03
286 +condo,region,07,K_qhpi_direct_local_td,0.39
287 +condo,region,07,S_sharp,0.0
288 +condo,region,07,S_sharp_filtered,0.0
289 +condo,region,12,A_median,0.01
290 +condo,region,12,B_stratified_median,0.02
291 +condo,region,12,C_pooled_td,0.1
292 +condo,region,12,D_rtd_w13_mean,4.55
293 +condo,region,12,E_hi_fisher_chained,0.65
294 +condo,region,12,F_geks_w25_movement,0.96
295 +condo,region,12,H_repeat_sales_cs,0.02
296 +condo,region,12,K_qhpi_hierarchical,0.01
297 +condo,region,12,S_sharp,0.0
298 +condo,region,12,S_sharp_filtered,0.0
299 +condo,region,13,A_median,0.0
300 +condo,region,13,B_stratified_median,0.01
301 +condo,region,13,C_pooled_td,0.12
302 +condo,region,13,D_rtd_w13_mean,5.36
303 +condo,region,13,E_hi_fisher_chained,5.08
304 +condo,region,13,F_geks_w25_movement,31.53
305 +condo,region,13,H_repeat_sales_cs,0.03
306 +condo,region,13,K_qhpi_direct_local_td,0.46
307 +condo,region,13,S_sharp,0.0
308 +condo,region,13,S_sharp_filtered,0.0
309 +condo,region,14,A_median,0.01
310 +condo,region,14,B_stratified_median,0.01
311 +condo,region,14,C_pooled_td,0.16
312 +condo,region,14,D_rtd_w13_mean,5.09
313 +condo,region,14,E_hi_fisher_chained,5.04
314 +condo,region,14,F_geks_w25_movement,32.05
315 +condo,region,14,H_repeat_sales_cs,0.02
316 +condo,region,14,K_qhpi_direct_local_td,0.51
317 +condo,region,14,S_sharp,0.0
318 +condo,region,14,S_sharp_filtered,0.0
319 +condo,region,15,A_median,0.01
320 +condo,region,15,B_stratified_median,0.02
321 +condo,region,15,C_pooled_td,0.14
322 +condo,region,15,D_rtd_w13_mean,5.66
323 +condo,region,15,E_hi_fisher_chained,5.19
324 +condo,region,15,F_geks_w25_movement,33.68
325 +condo,region,15,H_repeat_sales_cs,0.04
326 +condo,region,15,K_qhpi_direct_local_td,0.61
327 +condo,region,15,S_sharp,0.0
328 +condo,region,15,S_sharp_filtered,0.0
329 +condo,region,16,A_median,0.01
330 +condo,region,16,B_stratified_median,0.02
331 +condo,region,16,C_pooled_td,0.3
332 +condo,region,16,D_rtd_w13_mean,6.04
333 +condo,region,16,E_hi_fisher_chained,5.77
334 +condo,region,16,F_geks_w25_movement,34.3
335 +condo,region,16,H_repeat_sales_cs,0.05
336 +condo,region,16,K_qhpi_direct_local_td,0.74
337 +condo,region,16,S_sharp,0.0
338 +condo,region,16,S_sharp_filtered,0.0
339 +condo,region,17,A_median,0.0
340 +condo,region,17,B_stratified_median,0.01
341 +condo,region,17,C_pooled_td,0.08
342 +condo,region,17,D_rtd_w13_mean,4.46
343 +condo,region,17,E_hi_fisher,0.2
344 +condo,region,17,F_geks_w25,0.23
345 +condo,region,17,H_repeat_sales_cs,0.02
346 +condo,region,17,K_qhpi_hierarchical,0.01
347 +condo,region,17,S_sharp,0.0
348 +condo,region,17,S_sharp_filtered,0.0
349 +plex,province,QC,A_median,0.01
350 +plex,province,QC,A_mean,0.01
351 +plex,province,QC,B_stratified_median,0.02
352 +plex,province,QC,C_pooled_td,0.81
353 +plex,province,QC,C_pooled_td_ols,0.22
354 +plex,province,QC,D_rtd_w13_mean,9.8
355 +plex,province,QC,D_rtd_w13_movement,8.33
356 +plex,province,QC,D_rtd_w13_window,8.27
357 +plex,province,QC,D_rtd_w25_mean,10.4
358 +plex,province,QC,D_rtd_w7_mean,7.1
359 +plex,province,QC,E_hi_laspeyres_chained,6.79
360 +plex,province,QC,E_hi_paasche_chained,7.45
361 +plex,province,QC,E_hi_fisher_chained,7.02
362 +plex,province,QC,E_hi_fisher_direct,7.16
363 +plex,province,QC,F_geks_w25_movement,41.07
364 +plex,province,QC,F_geks_w13_movement,23.89
365 +plex,province,QC,H_repeat_sales_cs,0.11
366 +plex,province,QC,H_repeat_sales_bmn,0.11
367 +plex,province,QC,J_gbm_oot_residual,247.49
368 +plex,province,QC,K_qhpi,10.98
369 +plex,province,QC,S_sharp,19.12
370 +plex,province,QC,S_sharp_notv,20.43
371 +plex,region,01,A_median,0.01
372 +plex,region,01,B_stratified_median,0.02
373 +plex,region,01,C_pooled_td,0.17
374 +plex,region,01,D_rtd_w13_mean,5.62
375 +plex,region,01,E_hi_fisher_chained,0.4
376 +plex,region,01,F_geks_w25_movement,0.5
377 +plex,region,01,H_repeat_sales_cs,0.03
378 +plex,region,01,K_qhpi_hierarchical,0.04
379 +plex,region,01,S_sharp,0.0
380 +plex,region,01,S_sharp_filtered,0.0
381 +plex,region,02,A_median,0.01
382 +plex,region,02,B_stratified_median,0.01
383 +plex,region,02,C_pooled_td,0.16
384 +plex,region,02,D_rtd_w13_mean,5.8
385 +plex,region,02,E_hi_fisher_chained,5.34
386 +plex,region,02,F_geks_w25_movement,28.81
387 +plex,region,02,H_repeat_sales_cs,0.03
388 +plex,region,02,K_qhpi_direct_local_td,0.47
389 +plex,region,02,S_sharp,0.0
390 +plex,region,02,S_sharp_filtered,0.0
391 +plex,region,03,A_median,0.0
392 +plex,region,03,B_stratified_median,0.01
393 +plex,region,03,C_pooled_td,0.15
394 +plex,region,03,D_rtd_w13_mean,5.84
395 +plex,region,03,E_hi_fisher_chained,6.38
396 +plex,region,03,F_geks_w25_movement,38.19
397 +plex,region,03,H_repeat_sales_cs,0.03
398 +plex,region,03,K_qhpi_direct_local_td,0.59
399 +plex,region,03,S_sharp,0.0
400 +plex,region,03,S_sharp_filtered,0.0
401 +plex,region,04,A_median,0.0
402 +plex,region,04,B_stratified_median,0.03
403 +plex,region,04,C_pooled_td,0.13
404 +plex,region,04,D_rtd_w13_mean,5.46
405 +plex,region,04,E_hi_fisher_chained,5.68
406 +plex,region,04,F_geks_w25_movement,35.51
407 +plex,region,04,H_repeat_sales_cs,0.02
408 +plex,region,04,K_qhpi_direct_local_td,0.5
409 +plex,region,04,S_sharp,0.0
410 +plex,region,04,S_sharp_filtered,0.0
411 +plex,region,05,A_median,0.01
412 +plex,region,05,B_stratified_median,0.02
413 +plex,region,05,C_pooled_td,0.15
414 +plex,region,05,D_rtd_w13_mean,5.8
415 +plex,region,05,E_hi_fisher_chained,5.95
416 +plex,region,05,F_geks_w25_movement,35.68
417 +plex,region,05,H_repeat_sales_cs,0.03
418 +plex,region,05,K_qhpi_direct_local_td,0.51
419 +plex,region,05,S_sharp,0.0
420 +plex,region,05,S_sharp_filtered,0.0
421 +plex,region,06,A_median,0.01
422 +plex,region,06,B_stratified_median,0.01
423 +plex,region,06,C_pooled_td,0.27
424 +plex,region,06,D_rtd_w13_mean,7.09
425 +plex,region,06,E_hi_fisher_chained,6.79
426 +plex,region,06,F_geks_w25_movement,40.86
427 +plex,region,06,H_repeat_sales_cs,0.04
428 +plex,region,06,K_qhpi_direct_local_td,0.94
429 +plex,region,06,S_sharp,0.0
430 +plex,region,06,S_sharp_filtered,0.0
431 +plex,region,07,A_median,0.01
432 +plex,region,07,B_stratified_median,0.01
433 +plex,region,07,C_pooled_td,0.15
434 +plex,region,07,D_rtd_w13_mean,6.16
435 +plex,region,07,E_hi_fisher_chained,5.35
436 +plex,region,07,F_geks_w25_movement,32.29
437 +plex,region,07,H_repeat_sales_cs,0.02
438 +plex,region,07,K_qhpi_direct_local_td,0.54
439 +plex,region,07,S_sharp,0.0
440 +plex,region,07,S_sharp_filtered,0.0
441 +plex,region,08,A_median,0.01
442 +plex,region,08,B_stratified_median,0.01
443 +plex,region,08,C_pooled_td,0.14
444 +plex,region,08,D_rtd_w13_mean,6.43
445 +plex,region,08,E_hi_fisher_chained,0.42
446 +plex,region,08,F_geks_w25_movement,0.6
447 +plex,region,08,H_repeat_sales_cs,0.02
448 +plex,region,08,K_qhpi_hierarchical,0.02
449 +plex,region,08,S_sharp,0.0
450 +plex,region,08,S_sharp_filtered,0.0
451 +plex,region,09,A_median,0.0
452 +plex,region,09,B_stratified_median,0.01
453 +plex,region,09,C_pooled_td,0.1
454 +plex,region,09,D_rtd_w13_mean,5.29
455 +plex,region,09,E_hi_fisher,0.18
456 +plex,region,09,F_geks_w25,0.2
457 +plex,region,09,H_repeat_sales_cs,0.02
458 +plex,region,09,K_qhpi_hierarchical,0.01
459 +plex,region,09,S_sharp,0.0
460 +plex,region,09,S_sharp_filtered,0.0
461 +plex,region,11,A_median,0.0
462 +plex,region,11,B_stratified_median,0.01
463 +plex,region,11,C_pooled_td,0.1
464 +plex,region,11,D_rtd_w13_mean,5.23
465 +plex,region,11,E_hi_fisher,0.23
466 +plex,region,11,F_geks_w25,0.29
467 +plex,region,11,H_repeat_sales_cs,0.03
468 +plex,region,11,K_qhpi_hierarchical,0.02
469 +plex,region,11,S_sharp,0.0
470 +plex,region,11,S_sharp_filtered,0.0
471 +plex,region,12,A_median,0.0
472 +plex,region,12,B_stratified_median,0.01
473 +plex,region,12,C_pooled_td,0.15
474 +plex,region,12,D_rtd_w13_mean,5.96
475 +plex,region,12,E_hi_fisher_chained,5.53
476 +plex,region,12,F_geks_w25_movement,27.8
477 +plex,region,12,H_repeat_sales_cs,0.03
478 +plex,region,12,K_qhpi_direct_local_td,0.55
479 +plex,region,12,S_sharp,0.0
480 +plex,region,12,S_sharp_filtered,0.0
481 +plex,region,13,A_median,0.01
482 +plex,region,13,B_stratified_median,0.01
483 +plex,region,13,C_pooled_td,0.13
484 +plex,region,13,D_rtd_w13_mean,5.28
485 +plex,region,13,E_hi_fisher_chained,2.37
486 +plex,region,13,F_geks_w25_movement,8.82
487 +plex,region,13,H_repeat_sales_cs,0.02
488 +plex,region,13,K_qhpi_hierarchical,0.02
489 +plex,region,13,S_sharp,0.0
490 +plex,region,13,S_sharp_filtered,0.0
491 +plex,region,14,A_median,0.0
492 +plex,region,14,B_stratified_median,0.02
493 +plex,region,14,C_pooled_td,0.15
494 +plex,region,14,D_rtd_w13_mean,6.49
495 +plex,region,14,E_hi_fisher_chained,6.09
496 +plex,region,14,F_geks_w25_movement,36.87
497 +plex,region,14,H_repeat_sales_cs,0.02
498 +plex,region,14,K_qhpi_direct_local_td,0.48
499 +plex,region,14,S_sharp,0.0
500 +plex,region,14,S_sharp_filtered,0.0
501 +plex,region,15,A_median,0.0
502 +plex,region,15,B_stratified_median,0.01
503 +plex,region,15,C_pooled_td,0.18
504 +plex,region,15,D_rtd_w13_mean,6.55
505 +plex,region,15,E_hi_fisher_chained,8.13
506 +plex,region,15,F_geks_w25_movement,46.56
507 +plex,region,15,H_repeat_sales_cs,0.06
508 +plex,region,15,K_qhpi_direct_local_td,0.8
509 +plex,region,15,S_sharp,0.0
510 +plex,region,15,S_sharp_filtered,0.0
511 +plex,region,16,A_median,0.0
512 +plex,region,16,B_stratified_median,0.04
513 +plex,region,16,C_pooled_td,0.28
514 +plex,region,16,D_rtd_w13_mean,9.42
515 +plex,region,16,E_hi_fisher_chained,8.15
516 +plex,region,16,F_geks_w25_movement,46.26
517 +plex,region,16,H_repeat_sales_cs,0.05
518 +plex,region,16,K_qhpi_direct_local_td,1.02
519 +plex,region,16,S_sharp,0.0
520 +plex,region,16,S_sharp_filtered,0.0
521 +plex,region,17,A_median,0.01
522 +plex,region,17,B_stratified_median,0.02
523 +plex,region,17,C_pooled_td,0.18
524 +plex,region,17,D_rtd_w13_mean,6.9
525 +plex,region,17,E_hi_fisher_chained,2.31
526 +plex,region,17,F_geks_w25_movement,6.57
527 +plex,region,17,H_repeat_sales_cs,0.02
528 +plex,region,17,K_qhpi_hierarchical,0.02
529 +plex,region,17,S_sharp,0.0
530 +plex,region,17,S_sharp_filtered,0.0
added results/vintages/revision_policies_province.csv +4 −0
@@ -0,0 +1,4 @@
1 +class,months,mae_first_release_vs_final,bias_first_release_vs_final,mae_third_release_vs_final,rmse_first,rmse_third,mean_abs_true_change
2 +condo,52,0.0015620242064084064,0.00012999474800442522,0.0015485661521688215,0.0020789246210967667,0.0020302944477716685,0.007675301455052519
3 +house,52,0.0015532707391587841,-0.00012648911680931643,0.0015806828848220328,0.0020166096637717987,0.001978731246833613,0.015036741431184
4 +plex,52,0.0025609731874916717,-0.00013590920564457217,0.0025993659546795556,0.0035467549510441792,0.0035119400625493245,0.015285921387441144
added results/vintages/revisions_by_release.csv +109 −0
@@ -0,0 +1,109 @@
1 +class,level,release,obs,mae_rev_m1_vs_final,bias_rev_m1_vs_final,mae_rev_12m_vs_final,bias_rev_12m_vs_final,mae_rev_m1_vs_last_rt,mae_rev_12m_vs_last_rt,share_m1_rev_gt_1pct
2 +condo,municipality,1,3326,0.018218959210363525,0.000016688468351380448,0.035899729145987345,-0.00021023850922487306,0.010687310523249133,0.018543645860963492,0.539386650631389
3 +condo,municipality,2,3326,0.01830369137665762,0.0003420352663881496,0.035203232880077505,0.000014803896699086845,0.010651332141485633,0.01800571985494987,0.5426939266386049
4 +condo,municipality,3,3326,0.018253256323750244,0.00016979601424385027,0.034917315515444566,-0.00007206650223968797,0.010565016295267286,0.017769186486300787,0.5469031870114252
5 +condo,municipality,4,3326,0.018498001562029184,-0.00023268101188767977,0.03468796870069724,-0.00006589479040444394,0.010597045308313584,0.017713744598356427,0.5463018641010222
6 +condo,municipality,5,3326,0.01890243926725158,0.00021581986863248198,0.035178307015601325,-0.00016159941438881678,0.010842865379196508,0.017952100712717328,0.550210463018641
7 +condo,municipality,6,3326,0.01881462255308852,0.00028753995744244874,0.03532503607865267,-0.000701690943807063,0.01091535931184554,0.018263884216127575,0.5493084786530367
8 +condo,municipality,7,3326,0.018671778319748422,0.0003328626747500349,0.036041158805345176,-0.0008354496235605425,0.010687919474979822,0.018484741694745257,0.5472038484666266
9 +condo,municipality,8,3326,0.01853143947049999,-0.00016892757131686554,0.03586994538523937,-0.000523705132569303,0.010673162772855687,0.01829841904002192,0.544798556825015
10 +condo,municipality,9,3326,0.01858009111707426,-0.0003449297154774869,0.03545922010754932,-0.00018396812920626674,0.01050755377992387,0.017667555346076835,0.5514131088394468
11 +condo,municipality,10,3326,0.0185982825211036,-0.00032078107918533257,0.034781251007409114,0.00034764241879270803,0.01036657598267548,0.017125414330917934,0.5472038484666266
12 +condo,municipality,11,3326,0.018560632418666984,-0.00042042842032594265,0.034129410672541334,0.0011552891842648084,0.010121351342934837,0.016301168671367203,0.5532170775706554
13 +condo,municipality,12,3326,0.018713291616592024,-0.00008723296083991139,0.033784122050478343,0.001846917023178393,0.01008536273096997,0.016165427891360936,0.5544197233914612
14 +condo,province,1,52,0.0015620242064084064,0.00012999474800442522,0.0024936086388600643,0.002180781249067779,0.0015620242064083995,0.0024936086388600443,0.0
15 +condo,province,2,52,0.001554788940591851,0.00021421155617977366,0.0024556995539281204,0.0021945097556216576,0.0015547889405918415,0.002455699553928104,0.0
16 +condo,province,3,52,0.0016383411282475647,0.00007172615245495256,0.002385968263163059,0.002127974502410657,0.00163834112824756,0.0023859682631630474,0.0
17 +condo,province,4,52,0.0016851200317672967,0.00015760816287830706,0.002420257628965046,0.0021891023199823692,0.001685120031767297,0.0024202576289650305,0.0
18 +condo,province,5,52,0.0016060504415260231,0.0001648223290224312,0.0023661487983808183,0.002172771548220345,0.0016060504415260288,0.002366148798380805,0.0
19 +condo,province,6,52,0.0016567502022934756,0.00028747540064996235,0.002405214592967795,0.0022068356583684673,0.0016567502022934812,0.00240521459296778,0.0
20 +condo,province,7,52,0.0016786515058316567,0.00020647669810997852,0.0023240913353446534,0.0020709014067529876,0.0016786515058316586,0.002324091335344652,0.0
21 +condo,province,8,52,0.001655389863398211,0.00024328274921799384,0.0022743724214897825,0.0020328784287352046,0.0016553898633982099,0.002274372421489772,0.0
22 +condo,province,9,52,0.001684486401976774,0.0001661810791386928,0.0021740109318927232,0.0018353554819828383,0.0016844864019767622,0.002174010931892708,0.0
23 +condo,province,10,52,0.001593885852531655,0.00016859658055825427,0.002117012954797542,0.0018047875724924965,0.0015938858525316494,0.0021170129547975255,0.0
24 +condo,province,11,52,0.0016030353170648113,0.00016609797181298672,0.001982236262739768,0.0017272436940498302,0.0016030353170648011,0.001982236262739752,0.0
25 +condo,province,12,52,0.0015979690610675432,0.00020430782104002075,0.002030825256681998,0.0017622010900594173,0.0015979690610675526,0.002030825256681985,0.0
26 +condo,region,1,832,0.009790266447409655,-0.000031104480999510837,0.019273842101740615,-0.0034347501066969784,0.006077331017071481,0.011545513715423313,0.34134615384615385
27 +condo,region,2,832,0.009842418822069127,-0.00005817544655845735,0.01915099154614561,-0.003793582520574911,0.005901925322208566,0.011257364932724868,0.3401442307692308
28 +condo,region,3,832,0.009979287810595174,-0.00023492539186820973,0.01927819307081639,-0.004153633182955936,0.006011492519209637,0.011521417882571442,0.34134615384615385
29 +condo,region,4,832,0.010126823709202846,-0.00033397566967972305,0.019392051992608312,-0.004460492510153581,0.006075055890613615,0.01172819635641147,0.35216346153846156
30 +condo,region,5,832,0.010179366354186963,-0.0002865979063255834,0.019793882996838732,-0.0047910316429116825,0.006048436736559418,0.012013468400404903,0.36177884615384615
31 +condo,region,6,832,0.010367298598993564,-0.00022179320681323317,0.02002685451109216,-0.005158897833503317,0.006257140672470927,0.012170491393923113,0.3701923076923077
32 +condo,region,7,832,0.010192976572009942,-0.0001287222355680675,0.020767326355680602,-0.00459329080163186,0.006123962972048171,0.012422729646467559,0.36538461538461536
33 +condo,region,8,832,0.01020605005605612,-0.0004294500437435218,0.020638108116871796,-0.00445979331812907,0.006134809678767996,0.012179908602905022,0.35216346153846156
34 +condo,region,9,832,0.010106086957893909,-0.0004738868148322336,0.02058157780165701,-0.004521995638758401,0.006081644606522165,0.011882547908836187,0.34134615384615385
35 +condo,region,10,832,0.01008205418619355,-0.00033571880156921973,0.02034217407559042,-0.004484759347404402,0.005893884835961093,0.011589247146868181,0.34375
36 +condo,region,11,832,0.00999206557050614,-0.00041264004127453417,0.020295766446709166,-0.004603531125417201,0.005814847463540111,0.011335083931070078,0.3401442307692308
37 +condo,region,12,832,0.009847622079466455,-0.00048776006746468407,0.02003105454023685,-0.0044264189327692435,0.0056792893388342485,0.01112211780606469,0.33413461538461536
38 +house,municipality,1,3328,0.023092188875919885,-0.00010638684910900311,0.024208956837254495,-0.0009847268772388564,0.01611997040419284,0.017647401865411975,0.6538461538461539
39 +house,municipality,2,3328,0.02305826378044436,0.0000776618071806769,0.024121334462707294,-0.000733534018982744,0.016076172463857133,0.01754601518602517,0.6607572115384616
40 +house,municipality,3,3328,0.023314129313209635,-0.00018437102168174861,0.0243558017900071,-0.000805528223827719,0.016161186990447114,0.017631539281284643,0.6574519230769231
41 +house,municipality,4,3328,0.023094714275644447,-0.00001242905734893477,0.024438727072204902,-0.0007450148449550792,0.015984483975738917,0.017493185039469124,0.6487379807692307
42 +house,municipality,5,3328,0.02319062035455638,-0.00021508974361102599,0.024551891295953435,-0.0008033047850721947,0.016033372472342513,0.017595766690442386,0.6544471153846154
43 +house,municipality,6,3328,0.023469611454049084,0.00006959664267645442,0.024778345097935032,-0.0008921591898266268,0.016046905816510533,0.0178106438497968,0.6556490384615384
44 +house,municipality,7,3328,0.023757462631077236,-0.00008859729964317513,0.02580790419931541,-0.0010644822404534072,0.01616931963135506,0.017893589258424866,0.6526442307692307
45 +house,municipality,8,3328,0.02405027377329403,0.00003678475982662613,0.025928003607596735,-0.0009488613149150432,0.016375509629565892,0.017865497478796667,0.6589543269230769
46 +house,municipality,9,3328,0.024112394613001943,-0.0003514411949470708,0.025740121764867223,-0.0008296229136360721,0.016516847008696097,0.017721469248294538,0.6613581730769231
47 +house,municipality,10,3328,0.02415853926255957,-0.00005750907250976803,0.025534692908016896,-0.000630725627465175,0.016674363998256957,0.017462948328216068,0.6607572115384616
48 +house,municipality,11,3328,0.024346670361835077,-0.00006186481278868769,0.025607525987373706,-0.0005609351745473852,0.016998055871441813,0.017679407791658816,0.6613581730769231
49 +house,municipality,12,3328,0.02425204686382497,-0.00009108103528319962,0.025347260810956472,-0.00041864634387106785,0.016786336051117897,0.017276264088198927,0.6664663461538461
50 +house,province,1,52,0.0015532707391587841,-0.00012648911680931643,0.0012677792394390134,-0.0006682393886121419,0.0015532707391587958,0.0012677792394390347,0.0
51 +house,province,2,52,0.0016559392160083,-0.00008235723045537411,0.0013362779831777101,-0.0007866040235986188,0.0016559392160083047,0.0013362779831777225,0.0
52 +house,province,3,52,0.001583892533307463,-0.000043611870685160214,0.001412174589921807,-0.0008420760005469064,0.0015838925333074663,0.0014121745899218123,0.0
53 +house,province,4,52,0.0015542758198942297,-0.00009829276850291391,0.0015479190215160433,-0.001004210665083142,0.0015542758198942381,0.001547919021516063,0.0
54 +house,province,5,52,0.0014986613545432267,-0.00004873313943422304,0.0015508090247639411,-0.000989535425208228,0.0014986613545432334,0.0015508090247639593,0.0
55 +house,province,6,52,0.0015343173963489629,0.00006115591403258357,0.0015003670779485277,-0.0010266957469961696,0.0015343173963489594,0.0015003670779485405,0.0
56 +house,province,7,52,0.0015909958050343209,-0.0000190056585240687,0.0015965742541963228,-0.0011773220699620114,0.0015909958050343202,0.001596574254196325,0.0
57 +house,province,8,52,0.001452542351743374,-0.000034452325799709044,0.0015281850812553575,-0.0011388022641749306,0.0014525423517433717,0.0015281850812553608,0.0
58 +house,province,9,52,0.0014130734327221692,-0.0000909970886723188,0.0014631138909238816,-0.0011195581751700361,0.0014130734327221666,0.0014631138909238816,0.0
59 +house,province,10,52,0.0014782505022436498,-0.00007538162157755709,0.0015200536868807373,-0.0010272498521163638,0.0014782505022436407,0.0015200536868807328,0.0
60 +house,province,11,52,0.0015230332258875645,-2.250496928529023e-6,0.0014643526037635517,-0.0010100131367339448,0.0015230332258875609,0.0014643526037635562,0.0
61 +house,province,12,52,0.0015153200471414187,-0.00010782398525555524,0.0014670531390204471,-0.0009616150650127447,0.0015153200471414168,0.0014670531390204601,0.0
62 +house,region,1,884,0.019341166648931353,0.00013914401138195157,0.022435845036222188,-0.000054689108704933114,0.01783500175212258,0.021391270357664364,0.5305429864253394
63 +house,region,2,884,0.019461798425585455,-0.000020573850304829773,0.02195862296461838,-0.0001408973103168795,0.017848468990383434,0.020992692267059982,0.5271493212669683
64 +house,region,3,884,0.019640510488117734,0.0004008813035882496,0.022243928533564016,0.000037173789520607527,0.018068003224688576,0.021319531400553697,0.5373303167420814
65 +house,region,4,884,0.019351892777547952,0.00035417419660580137,0.02278751020208834,-0.0007848639223575325,0.017899705511420682,0.02177258496806936,0.5237556561085973
66 +house,region,5,884,0.019091097073210165,-0.00018818138959991127,0.022931771132624845,-0.0011419958694885563,0.017653390438943522,0.02165714183348216,0.5248868778280543
67 +house,region,6,884,0.019139962082195424,-0.00048142272931152904,0.023312582951215185,-0.0015064362333008757,0.017726656446890955,0.02189243380022784,0.5294117647058824
68 +house,region,7,884,0.01904469005779749,-0.0003719533571034606,0.02328476245583892,-0.000838248894445598,0.017692722186945373,0.021746034209333277,0.5418552036199095
69 +house,region,8,884,0.019261939850554546,-0.00026944596776266893,0.02329584417620687,-0.0004888714063228724,0.018005169160605173,0.02155822569409291,0.5373303167420814
70 +house,region,9,884,0.019267639722951157,-0.00003037649320443736,0.023305620403128734,-0.00035999914495570917,0.018186075121703563,0.021486607965907668,0.5350678733031674
71 +house,region,10,884,0.01914122368946062,0.00010720487911149635,0.023078835788251314,-0.00010840283347579008,0.01840151268136635,0.021350152425829293,0.5260180995475113
72 +house,region,11,884,0.01946217375677165,0.00017326483991043616,0.02332239123772778,0.00012918004126570213,0.01877205917976002,0.02168636582193896,0.5158371040723982
73 +house,region,12,884,0.019405316236140027,0.0001325954479839686,0.022926702846803357,0.00020440160720895,0.01880405105246462,0.02125108347601303,0.5090497737556561
74 +plex,municipality,1,3328,0.030288452162634486,-0.0005165186498657069,0.05262428846302046,-0.0013226885566379341,0.019897252981619543,0.033129705017398134,0.7349759615384616
75 +plex,municipality,2,3328,0.030266879188481352,-0.0003655847747851418,0.05124090476264386,-0.0008382047477597895,0.01989114441298446,0.032497132965224954,0.7409855769230769
76 +plex,municipality,3,3328,0.030331666170240246,-0.0002023008851125926,0.050822375628742804,-0.0013142855616835884,0.020014972187375853,0.032151556987150706,0.7400841346153846
77 +plex,municipality,4,3328,0.030477212650632968,0.00009434774566381735,0.05035652503326435,-0.001269901688993596,0.019986435679324968,0.031484862709542816,0.7484975961538461
78 +plex,municipality,5,3328,0.030384353808983886,-0.0002506971999339553,0.050557107407382555,-0.0015328881303532374,0.019852268026197755,0.03133718273027722,0.7563100961538461
79 +plex,municipality,6,3328,0.03047240597019304,-0.0006416287762726019,0.050801391274887295,-0.0012769798940341077,0.019520257572899293,0.03113832132401664,0.7478966346153846
80 +plex,municipality,7,3328,0.030097634848138103,0.0005177780192227456,0.05285719878650923,-0.00004594055955131114,0.019349417672062698,0.03183955846264396,0.7424879807692307
81 +plex,municipality,8,3328,0.029959285314090908,0.00012593807399265388,0.05247757118949334,-0.0007128007364661556,0.019118259744491658,0.03136403272927994,0.73828125
82 +plex,municipality,9,3328,0.030089488148415975,0.00045823265598124614,0.052224433356263526,-0.0009431994798718081,0.01901532501819489,0.03092802726606761,0.7391826923076923
83 +plex,municipality,10,3328,0.030194143109486052,-0.0003324314840843651,0.05221499953743071,-0.00144042255688976,0.019153754370377817,0.03092370755108541,0.7430889423076923
84 +plex,municipality,11,3328,0.030196699685274782,-0.0005071502270420179,0.05194727249336939,-0.0009267387182001881,0.019627275902071434,0.03058956960222138,0.7409855769230769
85 +plex,municipality,12,3328,0.030019016362902717,0.0002973269455979839,0.05150005703267959,-9.476544628591346e-6,0.019092770944119066,0.029870112776456893,0.7418870192307693
86 +plex,province,1,52,0.0025609731874916717,-0.00013590920564457217,0.0034534763426082977,-0.0015875438365012494,0.0025609731874916665,0.0034534763426082946,0.0
87 +plex,province,2,52,0.0024926043659790893,-0.00009454019214262746,0.0032054734636291324,-0.0014694410088973257,0.0024926043659790685,0.003205473463629128,0.019230769230769232
88 +plex,province,3,52,0.002553714357481464,-0.00004047794281379422,0.003094585608310371,-0.0015220777776656772,0.002553714357481438,0.003094585608310366,0.0
89 +plex,province,4,52,0.0027547193357477986,-0.00015209656244811245,0.003012014534139996,-0.0015026221629895072,0.0027547193357477804,0.0030120145341399926,0.0
90 +plex,province,5,52,0.0027800874380931634,-0.0002161250434366023,0.0029145807667974918,-0.0014055396426533068,0.0027800874380931396,0.002914580766797496,0.0
91 +plex,province,6,52,0.0026934949345003034,-0.0002022008710534018,0.0028829386532136163,-0.0012786384099770268,0.0026934949345002804,0.002882938653213614,0.0
92 +plex,province,7,52,0.002582845125215752,-0.00015788692632138522,0.0028886757938532334,-0.0013242037067489598,0.0025828451252157382,0.0028886757938532395,0.0
93 +plex,province,8,52,0.002639080129391382,-0.00021502241592675158,0.0027037215493154174,-0.0012176184392219242,0.002639080129391367,0.0027037215493154244,0.0
94 +plex,province,9,52,0.0025205385749526854,-0.00011984928174292413,0.002612417507449769,-0.001147538912493115,0.002520538574952671,0.0026124175074497737,0.0
95 +plex,province,10,52,0.002471275260724911,-0.00016212659033303835,0.0025179190218171587,-0.0009736164485255596,0.0024712752607249015,0.002517919021817157,0.0
96 +plex,province,11,52,0.0023786308212119735,0.00008397121580975361,0.002409360630624702,-0.0007613081176061588,0.0023786308212119666,0.002409360630624712,0.0
97 +plex,province,12,52,0.002372654388962432,-0.00017528002044779283,0.0025522750376679062,-0.0008278294224610996,0.0023726543889624327,0.0025522750376679223,0.0
98 +plex,region,1,884,0.013662550268308716,-0.0003106828386291632,0.020509090851500367,-0.004066840589696208,0.008588373597085176,0.013735934909154963,0.5328054298642534
99 +plex,region,2,884,0.01380938733905172,-0.00005507667807821284,0.020321558543735557,-0.004052497859157451,0.008743558386230163,0.013775239910577506,0.5395927601809954
100 +plex,region,3,884,0.013579203115088274,-0.00015589132251798307,0.02036969763369768,-0.004311733216030613,0.008691208480082676,0.013924305443190406,0.5248868778280543
101 +plex,region,4,884,0.013639439516273274,-0.00021329719406405447,0.020611391769852692,-0.0042585236155864474,0.008751234298494903,0.013840468072287814,0.5271493212669683
102 +plex,region,5,884,0.01382307529096038,-0.00031560954542261154,0.021284503765627488,-0.004218441495675318,0.008739481647510192,0.0142695085422924,0.5328054298642534
103 +plex,region,6,884,0.013543587939921909,-0.0007488494780056365,0.021803628079206022,-0.004280267650932594,0.008605126816542623,0.014021650843498165,0.5316742081447964
104 +plex,region,7,884,0.013396622009093124,-0.0005177203933074815,0.022461321433010415,-0.004277202804897456,0.008728996501200475,0.014374504850619765,0.5226244343891403
105 +plex,region,8,884,0.013425408159991752,-0.00042760211871508845,0.02204763031635865,-0.004091255775097806,0.00864160129784591,0.014318956170876553,0.5237556561085973
106 +plex,region,9,884,0.013700099938324965,-0.0002571476814299578,0.02202374907087782,-0.004011719013280068,0.008795275562324676,0.014197163527242035,0.5328054298642534
107 +plex,region,10,884,0.013954066485638647,-0.00038978431401202163,0.02195315127310212,-0.0037817512916404287,0.008956589306037753,0.014001053319166847,0.5452488687782805
108 +plex,region,11,884,0.014203977274158053,-0.00022764982825614356,0.02185184168913049,-0.0033312202054388597,0.009293019546997147,0.014028442873695878,0.5599547511312217
109 +plex,region,12,884,0.014069327265242619,-0.0004475291972578539,0.021825252254585842,-0.0032864627859346017,0.009161575036737011,0.014281809023605587,0.5497737556561086
added scripts/21_mc_summary.py +28 −0
@@ -0,0 +1,28 @@
1 +#!/usr/bin/env python3
2 +# Author: Simon-Pierre Boucher — simon-pierre.boucher@uqo.ca
3 +"""Phase 4 — NaN-safe Monte Carlo summary from results/simulation/metrics.parquet → summary.csv (means + Monte Carlo s.e.).
4 +Cells for which a method is not estimable (NaN) are excluded from that method's averages; `share_est` reports their share."""
5 +from __future__ import annotations
6 +
7 +import sys
8 +from pathlib import Path
9 +
10 +sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "src"))
11 +import polars as pl
12 +
13 +from sharp_hpi import config as C
14 +
15 +OUT = C.RESULTS / "simulation"
16 +m = pl.read_parquet(OUT / "metrics.parquet")
17 +cols = ["rmse_ret", "mae_ret", "bias_ret", "rmse_lvl", "cov95", "cov95_ret", "corr_ret", "lag_xcorr", "revision_rmse", "share_est", "secs"]
18 +m = m.with_columns([pl.col(c).fill_nan(None) for c in cols])
19 +# per replication first (so the Monte Carlo s.e. is over replications, not cells)
20 +rep = m.group_by(["scenario", "method", "level", "rep"]).agg([pl.col(c).mean().alias(c) for c in cols] + [pl.len().alias("cells")])
21 +summ = rep.group_by(["scenario", "method", "level"]).agg(
22 + pl.col("cells").mean().alias("cells"), pl.col("rep").n_unique().alias("reps"),
23 + *[pl.col(c).mean().alias(c) for c in cols],
24 + *[(pl.col(c).std() / pl.col(c).count().sqrt()).alias(c + "_mcse") for c in ["rmse_ret", "bias_ret", "rmse_lvl", "cov95"]],
25 +).sort(["scenario", "level", "rmse_ret"])
26 +summ.write_csv(OUT / "summary.csv")
27 +pl.Config.set_tbl_rows(200); pl.Config.set_tbl_width_chars(220)
28 +print(summ.filter(pl.col("level") == "municipality").select(["scenario", "method", "reps", "rmse_ret", "rmse_ret_mcse", "bias_ret", "cov95", "lag_xcorr", "share_est"]))
modified scripts/40_vintages.py +10 −7
@@ -2,7 +2,9 @@
2 2 # Author: Simon-Pierre Boucher — simon-pierre.boucher@uqo.ca
3 3 """Phase 6 — real-time vintages. For each monthly publication date v (10 days after month end, the measured publication
4 4 delay), re-estimate SHARP-HPI using only transactions dated ≤ v − 10 days (filtered mode: one-sided state-space deviations,
5 −stage 1 re-estimated on the available sample). Records every vintage of every cell path, then the revision statistics
5 +stage 1 re-estimated on the available sample). Records every vintage of every cell path in two modes — `filtered` (one-sided: the value of month m never
6 +changes after its first release except through stage-1 re-estimation; the no-revision policy) and `smoothed` (two-sided given the data
7 +available at the vintage: full re-estimation policy) — then the revision statistics
6 8 (1st/2nd/3rd release vs final), timeliness and three publication policies: (a) full re-estimation, (b) movement splice of
7 9 the newest month onto the frozen history, (c) revision-limited (history frozen after 3 releases).
8 10 Writes data/processed/index_vintages.parquet and results/vintages/*.csv."""
@@ -54,12 +56,13 @@ def main() -> None:
54 56 t0 = time.time()
55 57 r = run_sharp(avail, grid, realtime_parent=True)
56 58 n_prov = np.array([avail.filter(pl.col("month") == m).height for m in grid], float)
57 − rows.append(pl.DataFrame({"class": cls, "vintage": vm, "level": "province", "geo_id": "QC", "month": grid, "log_index": r.stage1.mu, "se": r.stage1.mu_se, "n": n_prov}))
58 − for (lvl, gid), c in r.cells.items():
59 − if lvl not in LEVELS_KEEP or (lvl == "municipality" and gid not in big_mun):
60 − continue
61 − p, v = cell_log_index(r, lvl, gid, "filtered")
62 − rows.append(pl.DataFrame({"class": cls, "vintage": vm, "level": lvl, "geo_id": gid, "month": grid, "log_index": p, "se": np.sqrt(v), "n": c.n}))
59 + for mode in ("filtered", "smoothed"):
60 + rows.append(pl.DataFrame({"class": cls, "vintage": vm, "mode": mode, "level": "province", "geo_id": "QC", "month": grid, "log_index": r.stage1.mu, "se": r.stage1.mu_se, "n": n_prov}))
61 + for (lvl, gid), c in r.cells.items():
62 + if lvl not in LEVELS_KEEP or (lvl == "municipality" and gid not in big_mun):
63 + continue
64 + p, v = cell_log_index(r, lvl, gid, mode)
65 + rows.append(pl.DataFrame({"class": cls, "vintage": vm, "mode": mode, "level": lvl, "geo_id": gid, "month": grid, "log_index": p, "se": np.sqrt(v), "n": c.n}))
63 66 print(cls, vm, f"{avail.height} tx, {time.time() - t0:.1f}s", flush=True)
64 67 pl.concat(rows).write_parquet(C.PROCESSED / "index_vintages.parquet")
65 68 print("done")
modified scripts/41_revisions.py +17 −5
@@ -21,8 +21,20 @@ OUT = C.RESULTS / "vintages"
21 21
22 22 def main() -> None:
23 23 v = pl.read_parquet(C.PROCESSED / "index_vintages.parquet")
24 − panel = pl.read_parquet(C.PROCESSED / "index_panel.parquet").select(
24 + modes = v["mode"].unique().to_list() if "mode" in v.columns else ["filtered"]
25 + if "mode" not in v.columns:
26 + v = v.with_columns(pl.lit("filtered").alias("mode"))
27 + for mode in modes:
28 + _analyse(v.filter(pl.col("mode") == mode), panel_path(), mode)
29 +
30 +
31 +def panel_path():
32 + return pl.read_parquet(C.PROCESSED / "index_panel.parquet").select(
25 33 pl.col("property_type").alias("class"), pl.col("geography_level").alias("level"), pl.col("geography_id").alias("geo_id"), "month", (pl.col("index_value") / 100).log().alias("li_final"))
34 +
35 +
36 +def _analyse(v: pl.DataFrame, panel: pl.DataFrame, mode: str) -> None:
37 + sfx = "" if mode == "smoothed" else "_" + mode
26 38 # release number of month m in vintage vm: k = months between m and vm (+1)
27 39 def mi(s):
28 40 return (s.str.slice(0, 4).cast(pl.Int32) * 12 + s.str.slice(5, 2).cast(pl.Int32))
@@ -43,7 +55,7 @@ def main() -> None:
43 55 pl.col("rev12_final").abs().mean().alias("mae_rev_12m_vs_final"), pl.col("rev12_final").mean().alias("bias_rev_12m_vs_final"),
44 56 pl.col("rev1_last").abs().mean().alias("mae_rev_m1_vs_last_rt"), pl.col("rev12_last").abs().mean().alias("mae_rev_12m_vs_last_rt"),
45 57 (pl.col("rev1_final").abs() > 0.01).mean().alias("share_m1_rev_gt_1pct")).sort(["class", "level", "release"])
46 − summ.write_csv(OUT / "revisions_by_release.csv")
58 + summ.write_csv(OUT / f"revisions_by_release{sfx}.csv")
47 59 # policies on the provincial monthly change: (a) full re-estimation = release k value; (b) movement splice = first release of each
48 60 # month chained; (c) revision-limited = value at release 3
49 61 pol = j.filter(pl.col("level") == "province")
@@ -55,9 +67,9 @@ def main() -> None:
55 67 (pl.col("d1_third") - pl.col("d1_final")).abs().mean().alias("mae_third_release_vs_final"),
56 68 ((pl.col("d1_first") - pl.col("d1_final")) ** 2).mean().sqrt().alias("rmse_first"), ((pl.col("d1_third") - pl.col("d1_final")) ** 2).mean().sqrt().alias("rmse_third"),
57 69 pl.col("d1_final").abs().mean().alias("mean_abs_true_change"))
58 − pol_summ.write_csv(OUT / "revision_policies_province.csv")
59 − j.filter(pl.col("release") <= 6).select(["class", "level", "geo_id", "month", "vintage", "release", "n", "log_index", "se", "d1", "d12", "rev1_final", "rev12_final"]).write_parquet(OUT / "revisions_long.parquet")
60 − print(summ); print(pol_summ)
70 + pol_summ.write_csv(OUT / f"revision_policies_province{sfx}.csv")
71 + j.filter(pl.col("release") <= 6).select(["class", "level", "geo_id", "month", "vintage", "release", "n", "log_index", "se", "d1", "d12", "rev1_final", "rev12_final"]).write_parquet(OUT / f"revisions_long{sfx}.parquet")
72 + print(mode); print(summ); print(pol_summ)
61 73
62 74
63 75 if __name__ == "__main__":
modified scripts/60_figures.py +13 −5
@@ -77,7 +77,7 @@ def fig_province_methods():
77 77 sub = s.filter((pl.col("class") == cls) & (pl.col("level") == "province") & (pl.col("method") == m)).sort("month")
78 78 if sub.height == 0:
79 79 continue
80 − months = sub["month"].to_list(); li = rebase(sub["log_index"].to_numpy(), months)
80 + sub = sub.filter(pl.col("month") < "2026-10"); months = sub["month"].to_list(); li = rebase(sub["log_index"].to_numpy(), months)
81 81 ax.plot(range(len(months)), li, color=METHOD_COLORS.get(m, "#999"), lw=1.6 if m == "S_sharp" else 1.0, label=METHOD_LABELS.get(m, m), alpha=1 if m == "S_sharp" else 0.9)
82 82 ax.set_title(CLASS_LABELS[cls]); month_ticks(ax, months); ax.axhline(100, color=INK2, lw=0.5)
83 83 axes[0].set_ylabel("Index (2021 average = 100)"); axes[0].legend(loc="upper left", fontsize=6.5)
@@ -97,11 +97,13 @@ def fig_region_methods():
97 97 sub = s.filter((pl.col("class") == "house") & (pl.col("level") == "region") & (pl.col("geo_id") == reg) & (pl.col("method") == m)).sort("month")
98 98 if sub.height == 0:
99 99 continue
100 − months = sub["month"].to_list(); li = rebase(sub["log_index"].to_numpy(), months)
100 + sub = sub.filter(pl.col("month") < "2026-10"); months = sub["month"].to_list(); li = rebase(sub["log_index"].to_numpy(), months)
101 101 col = METHOD_COLORS.get(m, METHOD_COLORS["K_qhpi"] if m.startswith("K_") else "#999")
102 + if m == "S_sharp_filtered":
103 + col = "#0b0b0b"
102 104 ax.plot(range(len(months)), li, color=col, lw=1.6 if m == "S_sharp" else 1.0, ls="--" if m == "S_sharp_filtered" else "-", label=METHOD_LABELS.get(m, "QHPI v2.1 architecture") if not m.startswith("K_") else "QHPI v2.1 architecture")
103 105 n = s.filter((pl.col("class") == "house") & (pl.col("level") == "region") & (pl.col("geo_id") == reg) & (pl.col("method") == "A_median")).sort("month")["n"].to_numpy()
104 − ax.set_title(f"{nm} — houses, median {np.nanmedian(n):.0f} sales/month"); month_ticks(ax, months); ax.axhline(100, color=INK2, lw=0.5)
106 + ax.set_title(f"{nm}\nhouses, median {np.nanmedian(n[:-1]):.0f} sales/month", fontsize=8); month_ticks(ax, months); ax.axhline(100, color=INK2, lw=0.5)
105 107 axes[0, 0].set_ylabel("Index (2021 = 100)"); axes[1, 0].set_ylabel("Index (2021 = 100)")
106 108 h, l = axes[0, 0].get_legend_handles_labels(); uniq = dict(zip(l, h)); fig.legend(uniq.values(), uniq.keys(), loc="lower center", ncol=4, fontsize=7.5, bbox_to_anchor=(0.5, -0.04))
107 109 save(fig, "fig_region_methods")
@@ -231,11 +233,17 @@ def fig_revisions():
231 233 return
232 234 d = pl.read_csv(p)
233 235 fig, axes = plt.subplots(1, 2, figsize=(8.4, 2.8))
236 + df_f = pl.read_csv(R / "vintages" / "revisions_by_release_filtered.csv") if (R / "vintages" / "revisions_by_release_filtered.csv").exists() else None
234 237 for i, lvl in enumerate(["province", "region", "municipality"]):
235 238 s = d.filter((pl.col("class") == "house") & (pl.col("level") == lvl)).sort("release")
236 − axes[0].plot(s["release"], 100 * s["mae_rev_m1_vs_final"], marker="o", ms=3, color=SERIES[i], label=lvl)
237 − axes[0].set_xlabel("Release number of a month"); axes[0].set_ylabel("Mean |revision| of monthly change vs final (pp)"); axes[0].legend(fontsize=7)
239 + axes[0].plot(s["release"], 100 * s["mae_rev_m1_vs_final"], marker="o", ms=3, color=SERIES[i], label=f"{lvl} (full re-estimation)")
240 + if df_f is not None:
241 + sf = df_f.filter((pl.col("class") == "house") & (pl.col("level") == lvl)).sort("release")
242 + axes[0].plot(sf["release"], 100 * sf["mae_rev_m1_vs_final"], ls="--", color=SERIES[i], lw=0.9, label=f"{lvl} (frozen first release)")
243 + axes[0].set_xlabel("Release number of a month"); axes[0].set_ylabel("Mean |revision| of monthly change vs final (log pts × 100)"); axes[0].legend(fontsize=6)
238 244 v = pl.read_parquet(q).filter((pl.col("class") == "house") & (pl.col("level") == "province"))
245 + if "mode" in v.columns:
246 + v = v.filter(pl.col("mode") == "smoothed")
239 247 vints = sorted(v["vintage"].unique().to_list())
240 248 for k, vm in enumerate(vints):
241 249 s = v.filter(pl.col("vintage") == vm).sort("month"); months = s["month"].to_list(); li = rebase(s["log_index"].to_numpy(), months)
modified scripts/61_tables.py +17 −2
@@ -252,7 +252,8 @@ def t_benchmarks():
252 252 d12c = c[12:] - c[:-12]; okk = ~np.isnan(d12a) & ~np.isnan(d12c)
253 253 if okk.sum() > 12:
254 254 vals_sh.append(np.sqrt(np.mean((d12a[okk] - d12c[okk]) ** 2)))
255 − r += [f"{100*np.mean(sds):.2f}" if sds else "—", f"{100*np.mean(vals_rs):.1f}" if vals_rs else "—"]
255 + r += [f"{100*np.nanmean(sds):.2f}" if sds else "—", f"{100*np.nanmean(vals_rs):.1f}" if vals_rs else "—"]
256 + macro(f"RegSd{m}{cls}", 100 * np.nanmean(sds) if sds else None, "{:.2f}"); macro(f"RegRmseRS{m}{cls}", 100 * np.nanmean(vals_rs) if vals_rs else None, "{:.1f}")
256 257 rows.append(r)
257 258 table("tab_benchmarks_region", ["Method", "H: s.d. $\\Delta$", "H: RMSE vs RS", "C: s.d. $\\Delta$", "C: RMSE vs RS", "P: s.d. $\\Delta$", "P: RMSE vs RS"], rows,
258 259 "Region-level benchmarks (16 administrative regions): mean standard deviation of monthly log changes (×100) and RMSE of 12-month log changes against the Case–Shiller repeat-sales index (×100)", "tab:bench_region", size="\\footnotesize",
@@ -327,7 +328,21 @@ def t_vintages():
327 328 p = pl.read_csv(R / "vintages" / "revision_policies_province.csv")
328 329 table("tab_policies", ["Class", "Months", "MAE first release", "Bias first release", "MAE third release", "RMSE first", "RMSE third", "Mean $|\\Delta|$"],
329 330 [[r["class"], str(r["months"]), f"{100*r['mae_first_release_vs_final']:.2f}", f"{100*r['bias_first_release_vs_final']:+.3f}", f"{100*r['mae_third_release_vs_final']:.2f}", f"{100*r['rmse_first']:.2f}", f"{100*r['rmse_third']:.2f}", f"{100*r['mean_abs_true_change']:.2f}"] for r in p.iter_rows(named=True)],
330 − "Provincial monthly change (×100): first and third release versus final", "tab:policies", notes="A movement-splice policy publishes the first-release change and never revises it; a revision-limited policy freezes the value at the third release.")
331 + "Provincial monthly change (×100): first and third release versus final", "tab:policies", notes="Full re-estimation policy (smoothed at each vintage). A revision-limited policy freezes the value at the third release; the frozen-first-release (movement-splice) policy is in Table~\\ref{tab:revisions_frozen}.")
332 + for r in p.iter_rows(named=True):
333 + macro(f"PolFirst{r['class']}", 100 * r["mae_first_release_vs_final"], "{:.2f}"); macro(f"PolThird{r['class']}", 100 * r["mae_third_release_vs_final"], "{:.2f}"); macro(f"PolMeanAbs{r['class']}", 100 * r["mean_abs_true_change"], "{:.2f}")
334 + if (R / "vintages" / "revisions_by_release_filtered.csv").exists():
335 + d = pl.read_csv(R / "vintages" / "revisions_by_release_filtered.csv")
336 + rows = []
337 + for cls in ["house", "condo", "plex"]:
338 + for lvl in ["province", "region", "municipality"]:
339 + x1 = d.filter((pl.col("class") == cls) & (pl.col("level") == lvl) & (pl.col("release") == 1)); x12 = d.filter((pl.col("class") == cls) & (pl.col("level") == lvl) & (pl.col("release") == 12))
340 + if x1.height:
341 + rows.append([f"{cls} / {lvl}", f"{100*x1['mae_rev_m1_vs_final'][0]:.2f}", f"{100*x12['mae_rev_m1_vs_final'][0]:.2f}" if x12.height else "—", f"{100*x1['mae_rev_12m_vs_final'][0]:.2f}", f"{100*x1['mae_rev_m1_vs_last_rt'][0]:.2f}"])
342 + macro(f"FrozenFirst{cls}{lvl}", 100 * x1["mae_rev_m1_vs_final"][0], "{:.2f}")
343 + table("tab_revisions_frozen", ["Class / level", "Rel. 1 vs final", "Rel. 12 vs final", "12-m change, rel. 1", "Rel. 1 vs last real-time"], rows,
344 + "Frozen-first-release (one-sided filtered) policy: mean absolute gap of the monthly log change (×100) between the real-time value and the final smoothed estimate", "tab:revisions_frozen", size="\\footnotesize",
345 + notes="Under this policy a month's value changes only through stage-1 re-estimation; the gap to the final series is the price of never revising (filter versus smoother), and it does not decline with later releases.")
331 346
332 347
333 348 def t_validation():
modified src/sharp_hpi/plotting.py +1 −1
@@ -10,7 +10,7 @@ SERIES = ["#2a78d6", "#eb6834", "#1baf7a", "#eda100", "#e87ba4", "#008300", "#62
10 10 INK = "#0b0b0b"; INK2 = "#52514e"; GRID = "#e4e3df"
11 11 METHOD_COLORS = { # fixed assignment: colour follows the entity
12 12 "S_sharp": SERIES[0], "C_pooled_td": SERIES[1], "A_median": SERIES[2], "H_repeat_sales_cs": SERIES[3], "D_rtd_w13_mean": SERIES[4],
13 − "E_hi_fisher_chained": SERIES[5], "F_geks_w25_movement": SERIES[6], "K_qhpi": SERIES[7], "G_hmts_w13": "#7a7a7a", "B_stratified_median": "#a0a0a0",
13 + "E_hi_fisher_chained": SERIES[5], "F_geks_w25_movement": SERIES[7], "K_qhpi": SERIES[6], "G_hmts_w13": "#7a7a7a", "B_stratified_median": "#a0a0a0",
14 14 }
15 15 METHOD_LABELS = {"S_sharp": "SHARP-HPI", "C_pooled_td": "Pooled time dummy (Huber)", "A_median": "Median price", "H_repeat_sales_cs": "Repeat sales (Case–Shiller)",
16 16 "D_rtd_w13_mean": "Rolling TD, 13 m, mean splice", "E_hi_fisher_chained": "Hedonic imputation Fisher (chained)", "F_geks_w25_movement": "GEKS-Fisher, 25 m",
17 17