Handling Fuzzy Spatial Data in R Using the fsr Package

Abstract

GIS and spatial data science (SDS) tools have been recently approaching each other by establishing bridge technologies between them. R as one of the most prominent programming languages used in SDS projects has been granted access to GIS infrastructure, while R scripts can be integrated and executed in GIS functions. Unfortunately, the treatment of spatial fuzziness has so far not been considered in SDS projects and bridge technologies due to a lack of software packages that can handle fuzzy spatial objects. This paper introduces an R package named fsr as an implementation of the fuzzy spatial data types, operations, and predicates of the Spatial Plateau Algebra that is based on the abstract Fuzzy Spatial Algebra. This R package solves the problem of constructing fuzzy spatial objects as spatial plateau objects from real datasets and describes how to conduct exploratory spatial data analysis by issuing geometric operations and topological predicates on fuzzy spatial objects. Further, fsr provides the possibility of designing fuzzy spatial inference models to discover new findings from fuzzy spatial objects. It optimizes the inference process by deploying the particle swarm optimization to obtain the point locations with the maximum or minimum inferred values that answer a specific user request.

fsr: An R package for fuzzy spatial data handling

“…This article extends the first version of fsr (introduced by Carniel et al (2021)) by: including the support for new data types that deal with heterogeneously structured fuzzy spatial collections and compositions (Carniel and Schneider, 2020); extending fuzzy geometric set operations and fuzzy numerical operations to handle all fuzzy spatial data types; specifying general operations that are based on fuzzy set theory and can be applied to any type of fuzzy spatial object; providing more details about the architecture of fsr and its implementation; and enhancing the running example and related programs to show the applicability of fsr using real datasets. …”

“…Throughout the remainder of this paper, we extend the application described by Carniel et al (2021) to illustrate how fsr works. We employ the following three real spatial datasets: accom_nyc_full : It consists of the Airbnb accommodations in New York City extracted from 2022‐03‐05 to 2022‐03‐31 ( http://insideairbnb.com/get‐the‐data.html); rest_nyc_full : It comprises the New York City restaurant inspection results provided by the Department of Health and Mental Hygiene (DOHMH) ( https://data.cityofnewyork.us/Health/DOHMH‐New‐York‐City‐Restaurant‐Inspection‐Results/43nn‐pn8j); and str_pav_nyc_full : It contains street pavement ratings in New York City provided by the New York City Department of Transportation ( https://data.cityofnewyork.us/Transportation/Street‐Pavement‐Rating/2cav‐chmn). …”

“…These operations are implemented in fsr by using the definitions provided by SPA. Finally, Fuzzy Spatial Inference Module provides a set of functions that allow users to create and evaluate fuzzy spatial inference models, as discussed by Carniel et al (2021, 2022). Table 2 briefly introduces the functions that are described in this article.…”

“… Note : The prefix “spa_” stands for Spatial Plateau Algebra. Note that the functions of the Fuzzy Spatial Inference Module are out of the scope of this article and more details on them can be found in Carniel et al (2021, 2022). …”