Can empirically based model results be fed into mathematical models? MCE for neural network and logistic regression in tourism landscape planning

Abstract

A wide variety of mathematical and empirical models have been implemented as practical tools for land-use planning, and multilayer perceptron (MLP), logistic regression or LR (mathematical model) and multi-criteria evaluation or MCE (empirical) are among widely applied models. One of the main drawbacks of the mathematical models is that they require dependent data and the process of data collection can be so costly and time-consuming for large areas. As such, we investigated the possibility of providing dependent data set through the MCE method for tourism planning in Golestan Province, Iran. The accuracy of MCE-based algorithms was investigated using ground truth data collected during field observations from early spring up to late summer 2016. The MCE-based and ground-based outputs were investigated and compared for spatial accuracy and connectivity and compactness of the results using receiving operator characteristic (ROC) and landscape configuration metrics. ROC statistics were scored at 0.886, 0.834, 0.82 and 0.814 for ground-based MLP, ground-based LR, MCE-based MLP and MCE-based LR, respectively, showing no meaningful differences between MCE-based and ground-based methods in terms of spatial accuracy. Landscape metrics also indicated that MCE-based methods have resulted in a more connected and manageable pattern for tourism planning. According to the results of this study, MCE can serve as a preliminary approach to define field sampling spots or even as an alternative to field observation efforts in case of limited time and financial resources.

Mapping Tourists’ Destination (Dis)Satisfaction Attributes with User-Generated Content

“…Advanced text-mining techniques could complement traditional methods [63]. In other words, adopting user-generated-content approaches has some benefits compared with traditional methods for data collection and data analysis [64]. This implies that this study might contribute to filling the absence of utility among tourism destinations and user-generated content.…”

Evaluating a guest satisfaction model through data mining

“…Adopting a data mining approach has several advantages when compared to other data analysis procedures. Specifically, data mining encompasses a set of modeling techniques grounded on machine learning and artificial intelligence which enable to apprehend complex (non-linear) relations between the variables, as opposed to traditional modeling techniques such as a logistic regression, which are more restrictive (Siroosi et al , 2019; Bloom, 2004). Furthermore, the difference in the predictive performance between a neural network model such as the one adopted in this study and a technique such as a logistic regression is amplified for larger data sets containing thousands of instances and several features, with a clear advantage in using a neural network (Eftekhar et al , 2005).…”

The sustainable transformation of business events: sociodemographic variables as determinants of attitudes towards sustainable academic conferences

“…), a computing method made up of processing units called “neurons” that function similarly to the human brain in terms of learning and storing knowledge. ANN has been successfully applied in studies of tourism (Fernández-Gámez et al , 2016; Siroosi et al , 2020) and consumer behaviour (Sakar et al , 2019), in which this technique outperformed traditional statistical techniques in terms of precision (Chong, 2013).…”