We develop an automated valuation model (AVM) for the residential real estate market by leveraging stacked generalization and a comparable market analysis. Specifically, we combine four novel ensemble learning methods with a repeat sales method and tailor the data selection for each value estimate. We calibrate and evaluate the model for the residential real estate market in Oslo by producing out-of-sample estimates for the value of 1,979 dwellings sold in the first quarter of 2018. Our novel approach of using stacked generalization achieves a median absolute percentage error of 5.4%, and more than 96% of the dwellings are estimated within 20% of their actual sales price. A comparison of the valuation accuracy of our AVM to that of the local estate agents in Oslo generally demonstrates its viability as a valuation tool. However, in stable market phases, the machine falls short of human capability.
%0 Journal Article
%1 birkeland2021predictability
%A Birkeland, Kristoffer B.
%A D’Silva, Allan D.
%A Füss, Roland
%A Oust, Are
%D 2021
%J International Real Estate Review
%K 2021 FDZ_GML SILC SILC_contra_zt SILC_input2021 article bat english jak text zt_proved
%N 2
%P 139-183
%T The Predictability of House Prices: “Human Against Machine”
%U https://www.gssinst.org/irer/2021/07/16/the-predictability-of-house-prices_human-against-machine/
%V 24
%X We develop an automated valuation model (AVM) for the residential real estate market by leveraging stacked generalization and a comparable market analysis. Specifically, we combine four novel ensemble learning methods with a repeat sales method and tailor the data selection for each value estimate. We calibrate and evaluate the model for the residential real estate market in Oslo by producing out-of-sample estimates for the value of 1,979 dwellings sold in the first quarter of 2018. Our novel approach of using stacked generalization achieves a median absolute percentage error of 5.4%, and more than 96% of the dwellings are estimated within 20% of their actual sales price. A comparison of the valuation accuracy of our AVM to that of the local estate agents in Oslo generally demonstrates its viability as a valuation tool. However, in stable market phases, the machine falls short of human capability.
@article{birkeland2021predictability,
abstract = {We develop an automated valuation model (AVM) for the residential real estate market by leveraging stacked generalization and a comparable market analysis. Specifically, we combine four novel ensemble learning methods with a repeat sales method and tailor the data selection for each value estimate. We calibrate and evaluate the model for the residential real estate market in Oslo by producing out-of-sample estimates for the value of 1,979 dwellings sold in the first quarter of 2018. Our novel approach of using stacked generalization achieves a median absolute percentage error of 5.4%, and more than 96% of the dwellings are estimated within 20% of their actual sales price. A comparison of the valuation accuracy of our AVM to that of the local estate agents in Oslo generally demonstrates its viability as a valuation tool. However, in stable market phases, the machine falls short of human capability.},
added-at = {2021-09-07T10:20:15.000+0200},
author = {Birkeland, Kristoffer B. and D’Silva, Allan D. and Füss, Roland and Oust, Are},
biburl = {https://www.bibsonomy.org/bibtex/26f087415824dcdb165c371a316a079c7/gesis_survey21},
description = {study
data-doi
gesis_study_no},
interhash = {1d8217a84a71a46278a56913d62c48db},
intrahash = {6f087415824dcdb165c371a316a079c7},
journal = {International Real Estate Review},
keywords = {2021 FDZ_GML SILC SILC_contra_zt SILC_input2021 article bat english jak text zt_proved},
language = {english},
note = {(SILC)},
number = 2,
pages = {139-183},
timestamp = {2022-01-11T11:57:57.000+0100},
title = {The Predictability of House Prices: “Human Against Machine”},
url = {https://www.gssinst.org/irer/2021/07/16/the-predictability-of-house-prices_human-against-machine/},
volume = 24,
year = 2021
}