3.1 Data cleaning and altered shop classification

3.1.1 Presentation of the data from RSV

3.1.1.1 The perimeter covered by the RSV database

Data have been provided by Rannsóknasetur verslunarinnar in an Excel Spreadsheet. Data includes the shop inventory from 2015, 2019 and June 2020 with the following variables:

• Ár
• Fyrirtæki
• Heimilisfang
• Numer
• Gata
• Meginflokkur
• Undirflokkur

It is worth noticing that the shop inventories do not cover the same perimeter over the three years (see the mapping of each year on Figure 3.1). Especially, the streets of Grandi are not included in the 2015 survey. No shop was identified on Njálsgata in 2015. The total number of shops in the database for the entire area is therefore not comparable. For sake of comparison, the statistics presented in this report do not include Grandi.

	2015	2019	2020
Grandagarður	0	73	108
Fiskislóð	0	72	68
Hafnartorg	0	4	28
Hólmaslóð	0	21	20
Garðastræti	0	16	14
Njálsgata	0	12	12
Eyjarslóð	0	11	11
Lindargata	0	6	6
Sölvhólsgata	0	4	5
Hallveigarstígur	0	4	3
Kárastígur	0	3	2
Veghúsastígur	0	1	1
Aðalstræti	0	1	0

Figure 3.1: Perimeter covered by the survey

3.1.1.2 The classification used by the RSV datase

The RSV database is based on a two-level classification:

• Category 1: Meginflokkur (3 categories in 2015; 4 categories in 2019 and 2020 - vacancy were not reported in 2015).
• Category 2: Undirflokkur (50 categories).

Figure 3.2: Shop classification used by RSV

As shown in the following table, in the original version of the database, ‘Önnur þjónustustarfsemi’ was the most represented category of shop (all year included) which can bias the interpretation of statistics. To avoid this pitfall, all the shops that were falling into this category have been reclassified in a better-suited category. Besides, several shops were not assigned any categories (#N/A or #REF).

Undirflokkur	Count	Percentage
Önnur þjónustustarfsemi	465	13.7
Matsölustaður	273	8.0
Föt og skór	263	7.8
Gistiheimili	176	5.2
Önnur verslun	162	4.8
Stofnanir/Samtök/Félög	129	3.8
Bar/Skemmtistaður	120	3.5
Arkitekt	114	3.4
Skyndibiti	112	3.3
Gjafa- og minjagripir	92	2.7
Lögfræðiþjónusta	85	2.5
Kaffi	80	2.4
Úr og skart	80	2.4
Ráðgjöf	72	2.1
Ráðuneiti/Embætti/svið	72	2.1
Ferðamannaþjónusta	71	2.1
Hárgreiðsla	71	2.1
#N/A	66	1.9
Hugbúnaðargerð	65	1.9
Önnur menning	65	1.9
	60	1.8
Listasafn	58	1.7
Auglýsingar og markaðsþjónusta	53	1.6
Fasteignir og leiga	46	1.4
Útgáfufélag	44	1.3
Matur	42	1.2
Kvikmynda- og tónlistagerð	40	1.2
Heilbrigðisþjónusta	39	1.1
Bækur, tónlist	37	1.1
Ljósmyndastofa	36	1.1
Sálfræðiaðstoð	35	1.0
Viðgerðir/Þrif	35	1.0
Hönnun	32	0.9
Búsáhöld	28	0.8
Gallerí	27	0.8
Tannlækningar	27	0.8
Önnur veitingastarfsemi	25	0.7
Snyrtivöruverslun	19	0.6
Leikhús / Kvikmyndir	18	0.5
Gleraugnaverslun	17	0.5
Fræðslusetur	14	0.4
Raftæki, tölvur	10	0.3
Hannyrðaverslun	9	0.3
Tómt	3	0.1
#REF!	2	0.1
Augnlæknir	1	0.0
Önnur þjónusta	1	0.0
Sálfræði aðstoð	1	0.0

3.1.2 Data cleaning : a new shop classification

The first step of this project has therefore consisted in a significant cleaning of the database to (1) complete missing data, (2) refine the shop classification and (3) reassign the proper category to most of the shops that were previously falling into the “Önnur þjónustustarfsemi” category. The websites https://ja.is and https://www.skatturinn.is/ have been used to allocate the appropriate category to these shops. Figure 3.3 and the following table show the new classification and the statistics after data cleaning.

The new shop classification is largely inspired by the one implemented by the Parisian planning agency (APUR). This institution has created an observatory of businesses and shops in early 2000s.¹ This observatory mainly relies on a periodical survey of all the shop in Paris, which fuels a georeferenced and open-access database of the retail structures in Paris. For each address, the survey mentions the name of the shop and its type, following a detailed three-level classification (see appendix).

Figure 3.3: New shop classification

Here are the new statistics after data cleaning:

Category 3	Count	Percentage
Restaurant	273	8.0
Clothes and shoes	263	7.8
Other service activities	194	5.7
Tourism accommodation	179	5.3
Institutions / Organizations / Associations	168	5.0
Other shop	162	4.8
Bar / Entertainment	122	3.6
Architect	117	3.4
Fast food	112	3.3
Vacancy	102	3.0
Advice	92	2.7
Gifts and souvenirs	92	2.7
Other cultural activity	92	2.7
Legal services	87	2.6
Tourist services	81	2.4
Café	80	2.4
Software development	80	2.4
Watches and jewelry	80	2.4
Ministry / Office / Division	72	2.1
Hairdressing	71	2.1
Advertising and marketing services	60	1.8
Art Museum	59	1.7
Real estate and rent	51	1.5
Publishing company	50	1.5
Film and music production	42	1.2
Food	42	1.2
Photography studio	40	1.2
Health care	39	1.1
Books, music	37	1.1
Repairs / Cleaning	37	1.1
Bank and financial service	36	1.1
Psychological assistance	36	1.1
Design	34	1.0
Household utensils	33	1.0
Other restaurant activities	33	1.0
Gallery	32	0.9
Dentistry	29	0.9
Other company service	23	0.7
Training center	20	0.6
Cosmetics store	19	0.6
Eyewear store	18	0.5
Other personal services	18	0.5
Theater / Movies	18	0.5
Electronics, computers	11	0.3
Gas station	9	0.3
Male hair shop	9	0.3
Gambling	8	0.2
Car rental	7	0.2
Car repair	6	0.2
Other medical practice	6	0.2
Beauty salon	5	0.1
Carwash	2	0.1
Other car-related company	2	0.1
Eye doctor	1	0.0
Taxi	1	0.0

3.2 Geocoding the database

Since the main purpose of the project is to provide a spatial analysis of shop changes in Miðborg, a key step was to geocode (i.e., to assign geographical coordinates) to each shop listed in the database in order to be able to map shops in a GIS. The geocoding process has been implemented with ArcGIS Pro (License: Sorbonne University, Faculté des Lettres). The database was split into three sub databases (one for 2015, one for 2019 and one for 2020), each of them was geocoded. A specific shapefile has been created for each sub-category of shop.

3.3 Mapping the georeferenced database onto an online mapping application

The geographical information produced through the geocoding process has been displayed on an online application available at that link: https://fl-su.maps.arcgis.com/apps/webappviewer/index.html?id=9dd7e923a6ca472fadabcf3f20d3617d

The application provides different tools:

• Display: Choose what layers you want to see on the map by clicking on the layer button .
• Explore:
  • To get more information about a peculiar shop or street, just click on the shop or street and a pop-up window will open.
  • You can use the select button to get some statistics about some subcategories of shop for a specific sub-area. Just draw on the map the area that you are interested in and a pop-up window will appear and the counting for the categories of shop you are interested in.
  • You can also display diagrams showing the number of shops for each category by clicking on the diagram button.
    
• Export: You can export the map as a *.pdf or other format by clicking on the print button.