Variable selection in STAR models with neighbourhood effects using genetic algorithms

Abstract

In this paper we deal with the problem of variable selection in spatiotemporal autoregressive (STAR) models with neighbourhood effects. We propose a procedure to carry out the selection process, taking into account the uncertainty associated with estimation of the parameters and the predictive behaviour of the compared models, in order to give more realism to the analysis. We set up a multi-objective programming problem that combines the use of different criteria to measure both these aspects. We use genetic algorithms which are very flexible and suitable for our multicriteria decision problem. In particular, the procedure allows us to estimate the number of spatial and temporal nearest neighbours as well as their relative effects. The methodology is illustrated through an application to the real estate market of Zaragoza. Copyright (C) 2010 John Wiley & Son, Ltd.STAR , variable selection , genetic algorithms , neighbourhood effects ,

Spatially varying coefficient models in real estate: Eigenvector spatial filtering and alternative approaches

“…Griffith (2008) originally proposed forward variable selection to find significant interactions, but this procedure is computationally slow (Seya et al, 2014) and only investigates iteratively a rather small number of variable combinations, posing the danger of an inappropriately selected set of variables. Because the model possibilities are 2 k , where k denotes the number of predictors, testing all possible models to determine the optimal combination computationally is rarely feasible (Alberto, Beamonte, Gargallo, Mateo, & Salvador, 2010). Additionally, simplified models are easier to interpret.…”

“…Stochastic search strategies, such as genetic algorithms (GA) (Goldberg, 1989;Reggiani, Nijkamp, & Sabella, 2001) imitating natural evolution, are effectively capable of selecting an optimal subset of covariates (e.g. Ahn et al, 2012;Alberto et al, 2010). Nevertheless, these approaches have so far been virtually ignored by real estate economists.…”

Putting time into space: the temporal coherence of spatial applications in the housing market

“…Along the same lines, Beamonte et al (2008Beamonte et al ( , 2010aBeamonte et al ( , 2010bBeamonte et al ( , 2013 and Alberto et al (2010) also employ Bayesian linear regressions, addressing heteroskedasticity problems, to estimate…”

“…Beamonte et al (2013) produce a constant quality housing index. Alberto et al (2010) developed a "multicriteria" methodology for variable selection and spatio-temporal effects in hedonic models. A genetic algorithm is proposed for identifying Pareto efficient solutions to solve the multi-objective problem.…”

A novel robust structural quadratic forecasting model and applications

“…The large amount of variables in massive datasets causes “curse of dimensionality” (COD), which poses unprecedented challenges and opportunities to forecasting and decision‐making system. To overcome COD, variable selection, which distinguishes important variables from noisy ones, is a popular and powerful technique in the realm of forecasting (Alberto et al, 2010; Hrdle et al, 2009; Refenes & Zapranis, 1999; Zeng, 2017). It can be regarded as a simplification procedure that extracts valuable data information from data by selecting important variable with predictive power (Ballings & Van den Poel, 2015; Bertsimas & Copenhaver, 2014; Ma et al, 2016; Wilms et al, 2016).…”