Methods for Regrouping Economic Activities into Meaningful Clusters

De Mulder, Sophie and Pennincx, Inge and Van Haute, Geert and Zaman, Jan (2021) Methods for Regrouping Economic Activities into Meaningful Clusters. CITIES 20.50 – Creating Habitats for the 3rd Millennium: Smart – Sustainable – Climate Neutral. Proceedings of REAL CORP 2021, 26th International Conference on Urban Development, Regional Planning and Information Society. pp. 499-508. ISSN 2521-3938

Text (Methods for Regrouping Economic Activities into Meaningful Clusters) CORP2021_78.pdf - Published Version Download (1MB)

Abstract

“The Flemish territory is characterized by a large urban sprawl […]. Even last years, an additional 6 hectares of undeveloped space is being built on daily. As a consequence open space is highly fragmented in Flanders“ (Pennincx, De Maeyer, Leroy, & De Mulder, 2021). As a strategic objective, the Flemish spatial government aims at a transition towards a net zero landtake daily by 2040. In this context, our spatial economy research group takes the choices and behaviour of individual companies and their use of space as a starting point. The main goal of the research is informing policy and supporting decision making by discerning spatial patterns, related to economic locations, and more precisely by focusing on the spatial environment of these locations. Over the years, we developed a the business-oriented approach for local spatial-economic policy and location advice for companies (Giaretta, Zaman, Pennincx, & De Mulder, 2019; Zaman, Pennincx, & De Mulder, 2020). For this, we need the exact location of the activity and the exact activity of every economic site. However,this information is difficult to gather from the only area-wide economic administrative database for the whole territory of Flanders (VKBO) (Gruijthuijsen et al., 2018). This area-covering database is used for major spatial-economic analyses, but it falls short in precision at the detail level needed for our work. We have carried out quite a lot of research in recent years to get to know the terrain situation by creating a field inventory. A key element of the research is the search for the right spatial synthesis of the data collected at the level of the parcel: through economic ecotopes and market segments we sought to combine the (economic) parcels into meaningful groups with similar characteristics. We described this step in previous papers (Giaretta et al., 2019; Zaman et al., 2020). Although the past research is interesting for the local policy makers of the mapped area’s, we still need to find a way to also make meaningful statements on spatial economic patterns for other areas in Flanders that have not been mapped. Producing this area-covering map for Flanders is rather important, as it will enable us to translate the analyses and the knowlegde we have gathered to (regional) policy. Although being thourough and rather precise, the visual inventory method has some drawbacks: it is time consuming and at this point, it cannot be easily applied to the entire area of Flanders. We therefore opt to first assess if we can extract useful statements regarding economic patterns from administrative databases. The main research question is whether the synthesis of the mapping data into the economic ecosystems or economic segments can be reproduced with the administrative database. Obviously, the results from the administrative database and the inventory will not be 100% alike. However, we believe it is possible come to spatial economic meaningful groups, even using the administrative database. The purpose of this grouping remains the same as with the inventory work and economic ecotopes and segments: being able to inform policy choices related to economic locations. In a first step, we examined whether and how the area synthesis (starting from the inventory and resulting into economic ecotopes and segments), that was carried out with manual work, field knowledge and expert opinion can be reproduced through automated methods, specifically through (1) statistical approach and/or machine learning and (2) a spatial predefined spatial clustering. The automated grouping results are reviewed and spatially analysed by spatial planners with territory knowledge. Only in a second step, when the grouping results on basis of the inventory are satisfying, we will rerun the method with the administrative data of the VKBO. In this paper we will discuss the first few steps of the grouping methods, in particular the distance and the activities clustering. We will outline the next steps, using the VKBO-data, assessing if we can come to meaningful economic clusters.

Item Type:	Article
Uncontrolled Keywords:	machine learning, clustering, Spatial economic patterns, Business Perspective, Flanders
Subjects:	H Social Sciences > HB Economic Theory H Social Sciences > HD Industries. Land use. Labor H Social Sciences > HM Sociology
Depositing User:	REAL CORP Administrator
Date Deposited:	27 Sep 2021 13:21
Last Modified:	17 Oct 2021 17:28
URI:	http://repository.corp.at/id/eprint/778

Actions (login required)

View Item