Predicting owner-occupied housing values using machine learning: an empirical investigation of California census tracts data

Simlai, Prodosh

doi:10.1080/09599916.2021.1890187

Cited by 5 publications

(4 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…En el ámbito específico de los modelos heredados de la inteligencia artificial, Tchuente y Nyawa (2022) concluyen sobre una muestra de diferentes ciudades francesas y para un periodo de 5 años que las redes neuronales artificiales y el modelo random forest superan de forma significativa a otros métodos cuando no se tienen en cuenta las características de geocodificación de los inmuebles, mientras que los modelos adaboost, gradient boosting y el propio random forest funcionan mejor cuando las características de geocodificación son incluidas en la muestra de datos. Simlai (2021) analiza el mercado inmobiliario de California, concluyendo que los métodos de inteligencia artificial proporcionan una descripción exhaustiva de los determinantes del valor de las viviendas en el conjunto de secciones censales de California. En comparación con los modelos hedónicos, las regresiones Ridge, LASSO y Elastic Net proporcionan predicciones fuera de muestra significativamente mejores.…”

Section: Revisión De La Literaturaunclassified

Valoración automática de inmuebles residenciales mediante modelos de Machine Learning

Guijarro¹

2023

ReE

View full text Add to dashboard Cite

La literatura reciente sobre valoración inmobiliaria ha aportado diversas evidencias en el ámbito internacional sobre el buen desempeño de los modelos de machine learning en la predicción del comportamiento de los precios, sobre todo si se comparan con los obtenidos por los denominados métodos tradicionales valoración, muy extendidos sobre todo en la práctica profesional. Con todo, se siguen remarcando algunas limitaciones como el diseño de caja negra y la dificultad en la interpretación de los resultados proporcionados por estas técnicas. Este trabajo tiene por objeto comparar los resultados y el desempeño de diferentes modelos de machine learning aplicados en el ámbito de la valoración inmobiliaria residencial. Para ello se ha recopilado una amplia base de datos con ofertas de inmuebles en la ciudad de Madrid, que permite dividir la muestra en los grupos de entrenamiento y test. La comparación entre los modelos se ha llevado a cabo a través de diferentes métricas, entre las que destaca el MAPE (Mean Absolute Percentage Error) por ser uno de los preferidos por las sociedades de tasación. Las métricas utilizadas confirman un buen rendimiento generalizado para el conjunto de modelos entrenados, con variaciones relativamente pequeñas tras el proceso de validación.

show abstract

Section: Revisión De La Literaturaunclassified

Valoración automática de inmuebles residenciales mediante modelos de Machine Learning

Guijarro¹

2023

ReE

View full text Add to dashboard Cite

show abstract

“…Simultaneously, its strong generalization ability, fitting accuracy, and nonlinear mapping ability make it more favorable among scholars in the real estate appraisal field [105,107]. In terms of accuracy, several studies have also shown that machine learning (ML) regression has better out-of-sample predictive ability compared with OLS regression [107,108]. The most commonly used methods of machine learning algorithms in the field include artificial neural networks (ANNs), SVM, deep learning (DL), and ensemble learning (EL) [85,105,107].…”

Section: Machine Learningmentioning

confidence: 99%

The Research Development of Hedonic Price Model-Based Real Estate Appraisal in the Era of Big Data

Wei

Wang

et al. 2022

Land

View full text Add to dashboard Cite

In the era of big data, advances in relevant technologies are profoundly impacting the field of real estate appraisal. Many scholars regard the integration of big data technology as an inevitable future trend in the real estate appraisal industry. In this paper, we summarize 124 studies investigating the use of big data technology to optimize real estate appraisal through the hedonic price model (HPM). We also list a variety of big data resources and key methods widely used in the real estate appraisal field. On this basis, the development of real estate appraisal moving forward is analyzed. The results obtained in the current studies are as follows: First, the big data resources currently applied to real estate appraisal include more than a dozen big data types from three data sources; the internet, remote sensing, and the Internet of things (IoT). Additionally, it was determined that web crawler technology represents the most important data acquisition method. Second, methods such as data pre-processing, spatial modeling, Geographic information system (GIS) spatial analysis, and the evolving machine learning methods with higher valuation accuracy were successfully introduced into the HPM due to the features of real estate big data. Finally, although the application of big data has greatly expanded the amount of available data and feature dimensions, this has caused a new problem: uneven data quality. Uneven data quality can reduce the accuracy of appraisal results, and, to date, insufficient attention has been paid to this issue. Future research should pay greater attention to the data integration of multi-source big data and absorb the applications developed in other disciplines. It is also important to combine various methods to form a new united evaluation model based on taking advantage of, and avoiding shortcomings to compensate for, the mechanism defects of a single model.

show abstract

“…That is, standard parametric econometric approaches will reveal critical pitfalls to examine such complex relations due to their pre-built, rigid functional forms. As a result, novel approaches such as various machine learning (ML) techniques have become promising alternatives thanks to their flexible, evolving features to predict market dynamics (see e.g., Ho et al, 2021;Pérez-Rave et al, 2019;Simlai, 2021).…”

Section: Introductionmentioning

confidence: 99%

Does Machine Learning Prediction Dampen the Information Asymmetry for Non-Local Investors?

Jung

Kim

Jin

2022

International Journal of Strategic Property Management

View full text Add to dashboard Cite

In this study, we examine the prediction accuracy of machine learning methods to estimate commercial real estate transaction prices. Using machine learning methods, including Random Forest (RF), Gradient Boosting Machine (GBM), Support Vector Machine (SVM), and Deep Neural Networks (DNN), we estimate the commercial real estate transaction price by comparing relative prediction accuracy. Data consist of 19,640 transaction-based office properties provided by Costar corresponding to the 2004–2017 period for 10 major U.S. CMSA (Consolidated Metropolitan Statistical Area). We conduct each machine learning method and compare the performance to identify a critical determinant model for each office market. Furthermore, we depict a partial dependence plot (PD) to verify the impact of research variables on predicted commercial office property value. In general, we expect that results from machine learning will provide a set of critical determinants to commercial office price with more predictive power overcoming the limitation of the traditional valuation model. The result for 10 CMSA will provide critical implications for the out-of-state investors to understand regional commercial real estate market.

show abstract

Predicting owner-occupied housing values using machine learning: an empirical investigation of California census tracts data

Cited by 5 publications

References 41 publications

Valoración automática de inmuebles residenciales mediante modelos de Machine Learning

Valoración automática de inmuebles residenciales mediante modelos de Machine Learning

The Research Development of Hedonic Price Model-Based Real Estate Appraisal in the Era of Big Data

Does Machine Learning Prediction Dampen the Information Asymmetry for Non-Local Investors?

Contact Info

Product

Resources

About