Mapping methodologies in Fabrix

Rotterdam and Athens are different cities; one a West-European port city, and the other a South-European capital city, and neither are the first you would think of when you think of fashion or textiles. Yet, these cities were chosen for their different relationship to textiles. So then how did we map such distinct cities to be able to see how their relationship to textiles are embedded in space? And how can you do the same for a city of your choosing? 

For the research into the business distribution in the cities, we used two main programmes: R / Rstudio, and QGIS. Both of these programmes are open-source alternatives to Python and ArcGIS respectively, and either can be used when following our process, though R / Rstudio and QGIS are the easiest to grasp for beginners, and there are many online forums you can consult whenever you run into issues when using them. 

Gathering Data 

For our research into Rotterdam, we used the LISA (Landelijke Informatiesysteem van Arbeidsplaatsen; National Employment Information System) dataset. The convenient part about this dataset is that there is a file available that logs every business in the Netherlands for every year since 1996. For our analysis, we used the timespan between 1996-2022. 

For Athens, we did not find a comparable dataset available for every year. Instead, we used the GeMI website to collect all the currently active businesses available for the latest year, 2023 (during the time of our analysis). 

Filtering the Data 

To map out the textile and clothing sector, we first have to define what this sector is comprised of. Using the document ‘Mode in Rotterdam’ (Fashion in Rotterdam) as a reference, we defined 11 key business categories that make up the textiles sector. These include manufacturers (of textiles), suppliers (manufacturers of parts), retailers (divided into first-hand and second-hand), maintenance actors (cleaning and repair), technology-related actors (in relation to sewing machines), designers, educational institutions, and others. These categories were created based on the different SBI codes (Standaard Bedrijfsindeling, Standard Industrial Classification – the Dutch equivalent of the European Statistical Classification of Economic Activities (NACE) codes) that relate to the textile and clothing sector.  

Category	NACE codes
Manufacturing of textiles	1411 ; 1412 ; 1413 ; 1414 ; 1419 ; 1420 ; 1431 ; 1439 ; 1511 ; 1512 ; 1520 ; 2060
Manufacturing of Parts	1310 ; 1320 ; 1330 ; 1391 ; 1392 ; 1393 ; 1394 ; 1395 ; 1396 ; 1399
Wholesale	4611 ; 4616 ; 46241 ; 46242 ; 46411 ; 46412 ; 46421 ; 46422 ; 46423 ; 46424 ; 46425 ; 46429 ; 46761
Retail	47511 ; 47512 ; 47513 ; 47711 ; 47712 ; 47713 ; 47714 ; 47715 ; 47716 ; 47717 ; 47718 ; 47721 ; 47722
Second-hand Retail	47792 ; 4782 ; 77292
Maintenance (Repair and Cleaning)	9523 ; 9529 ; 96011 ; 96012 ; 96013
Waste Collection	3832
Designers	74102
Network	8230
Education	85522
Technology (Sewing Machines)	2894 ; 4664 ; 47542

To filter the data to the scope of the textile and clothing sector in Rotterdam, we loaded the dataset (in .csv or .rds file formats) into Rstudio. After defining the different categories of SBI codes, the data files can be filtered to select only the businesses that belong to each business category. These files are exported to .csv files to then be visualised in QGIS. This process is repeated for every year of available data. 

Visualising Maps in QGIS 

The already-filtered exported data files can now be loaded into QGIS to visualise. Since each data file already contains only the businesses of a certain category, all QGIS is used for is to understand their spatial distribution. By adjusting the colours of the points and changing the type of points to heatmap, the specific concentrations of businesses can be more easily seen. When combined together, you can see where different business categories are located, and find trends or patterns in where they cluster. 

Your own application

You may have already collected your own dataset to find hotspots through. The most important information you need from it are the geographical coordinates of the businesses you’re considering and the NACE codes of business activities (as detailed as possible). Once you have these two, you can go wild with finding the spatial distribution of all kinds of businesses. Additional information like profit, employment, and gender work to enrich the data further, but with just location and activity, you can already get really far in your spatial analysis, particularly when combined with qualitative research into the neighbourhoods.