@article{ab6879e49ebf4d7cb0b7e7426812e5ef,
title = "Developing land use regression models for environmental science research using the XLUR tool – More than a one-trick pony",
abstract = "Land use regression (LUR) is a widely used method to develop prediction models in environmental sciences. However, the process of creating and applying LUR models is repetitive and time-consuming. The XLUR tool was developed to automate this process, while at the same time providing a detailed log of the model building process for reproducibility, and providing evaluation metrics to assess model quality. The aim of this research is to provide a technical demonstration of the use of XLUR in two scenarios.We demonstrate the use of the XLUR tool to build models for predicting PM10 concentrations in Greater Manchester and intestinal enterococci along the Northwest coast of England. The examples show how the tool facilitates (a) model building using standard published protocols and (b) assessment of prediction quality. As is common with LUR approaches, prediction quality is reliant on data and the characteristics of the phenomena being modelled.",
keywords = "GIS, Hybrid, Land use regression, Spatial data, Wizard",
author = "Anna M{\"o}lter and Sarah Lindley",
note = "Funding Information: The XLUR tool (Molter, 2020) was developed to automate the process of creating and applying LUR models for an ongoing air pollution study in Indonesia. It provides a wizard style interface to calculate and extract potential predictor variables, which simplifies step 2 described above. Importantly, the tool facilitates the generation of predictor variables from input data, saving time and also reducing the GIS skills required to carry out preparatory spatial analysis. This means that the method can be used by a range of environmental specialists who are not necessarily GIS specialists or statisticians. XLUR largely automates steps 3 and 4, thereby reducing the potential for human error and saving time. The XLUR tool uses the ESCAPE methodology; however, since LUR can be used to model a range of environmental processes, XLUR was designed as a general purpose tool, not solely for air pollution or air pollution epidemiological research. This means it provides more options for the extraction of predictor variables than a standard air pollution LUR tool would, but it still uses the robust supervised machine learning process set out in the ESCAPE manual. Furthermore the tool also facilitates the inclusion of additional model outputs as predictor variables, thus supporting the development of hybrid LUR models which were not part of the original ESCAPE methodology (De Hoogh et al., 2016).This work is supported via the NERC Newton-DIPI Urban hybriD models for AiR pollution exposure (UDARA) project, PIs: Prof G McFiggans, Faculty of Science and Engineering, The University of Manchester, UK, and Dr D Driejana, Faculty of Civil and Environmental Engineering, Institut Teknologi Bandung, Indonesia, NE/P014631/1. It builds on work carried out in the European Union's Seventh Framework Programme Theme ENV. 2007.1.2.2.2. European cohort on air pollution. Publisher Copyright: {\textcopyright} 2021 The Author(s)",
year = "2021",
month = sep,
day = "1",
doi = "10.1016/j.envsoft.2021.105108",
language = "English",
volume = "143",
journal = "Environmental Modelling & Software",
issn = "1364-8152",
publisher = "Elsevier BV",
}