In November 2018, I wrote a blog<\/SPAN><\/A> about an analysis on the BAG to distinguish housing types. There were many responses to this, and therefore we decided as a content team to dive deeper into this. And I am proud to announce that our new dataset Woningtypering is now available for our customers as a feature service<\/SPAN><\/A> and webmap<\/SPAN><\/A><\/SPAN> in ArcGIS Online! But of course, some work preceded this. I would like to take you through our approach. A complete description of all our design choices can be found in the Product Information. This can be downloaded from our website<\/A>. <\/SPAN><\/P> <\/SPAN><\/P>In my previous blog, I had already discovered that by using the tools Polygon Neighbors<\/SPAN><\/A> and Frequency<\/SPAN><\/A>, many of the housing types can be identified. The Polygon Neighbors tool finds adjacent polygons, in this case the adjacent neighboring buildings, and the Frequency tool counts these. We also made extensive use of these tools for this service. The Dissolve<\/SPAN><\/A> tool ultimately proved to be the solution for our "corner house versus semi-detached house problem." Because how do you know if a house is a corner house or a semi-detached house? Both types of houses naturally have only one neighbor. That neighbor is key here. Because either this neighbor itself has only one neighbor (the original building), which in that case is a semi-detached house, or this neighbor has multiple neighbors itself, after which the original building can be classified as a corner house.<\/SPAN><\/P> <\/SPAN><\/P>The simplest way to find out this information was to perform a Dissolve. Here, all adjoining buildings are connected into one polygon. At first, we thought of dissolving everything, where multipart was not allowed, and then checking whether buildings with only one neighbor were in a Dissolve consisting of two or multiple buildings. With the help of a Spatial Join<\/SPAN><\/A>, you can find out how many buildings lie within the dissolved polygon. A house with only one neighbor but in a polygon with multiple buildings would then be classified as a corner house in this Dissolve. Two semi-detached houses would only be dissolved with each other since they have no other neighbors. From this, we could conclude that if a house with only one neighbor is in a Dissolve consisting of two houses, it is a semi-detached house.<\/SPAN><\/P><\/SPAN><\/P>But it turned out not to be so "simple." After testing, we encountered some problems. For example, garages were often incorrectly designated as neighbors. According to the technical analysis, that was logical but not a desired outcome. Additionally, we found out that not all buildings were neatly drawn next to each other. For finding neighbors, this was not a problem since we could set a search margin of three centimeters which solved most issues. But for a Dissolve, this was not possible without much extra processing time. This could result in a situation where a building was considered a neighbor but was not included in the Dissolve. We also discovered that multiple types of buildings were designated as semi-detached houses, for example an apartment complex and a corner house located next to each other.<\/SPAN><\/P> <\/SPAN><\/P>To prevent all these situations, we adjusted our approach. First, we removed all buildings with only one single residential unit with usage purpose "overig" from our analysis. Buildings with one residential unit with usage purpose "overig" are often garages and this way we could solve our first problem. Then we applied the Polygon Neighbors tool again. Based on this result, we selected all buildings that had only one adjacent building. From those, we made a sub-selection of buildings with only one single residential unit. This way we prevented houses next to apartments from being designated as semi-detached houses. Then we performed a Dissolve only on these buildings with just one adjacent neighbor and only one residential unit (which did not have usage purpose "overig"). If the building was then in a Dissolve consisting of only one building (itself), it was a corner house because its neighbors themselves had multiple adjacent buildings or multiple residential units and were therefore not included in the Dissolve. If the building was then in a Dissolve consisting of two buildings, then the building was a semi-detached house. In this way, we also solved the problem of non-adjacent buildings and shortened our processing time because fewer buildings needed to be dissolved.<\/SPAN>
<\/SPAN><\/P>In my previous blog, I had already discovered that by using the tools
Of course, we made many more design choices. For example, we chose to include all buildings with residential units in our analysis because the use of the adjacent building is less important (except for garages). In our opinion, a house between two shops is an intermediate house and not detached property. From this reasoning also followed our decision to designate buildings with multiple residential units as apartments, provided that at least one of those residential units had usage purpose "wonen." A house above a shop, for example, is in our view an apartment.<\/SPAN>
You can find these and all our other design choices in our Product Information Woningtypering. This document contains information about the analysis process and design choices with examples of the consequences these choices have had regarding the outcome of the dataset. You can download this product information via our website<\/_a>. <\/_span>
<\/_p>
As a customer, you have free access from today to our feature service<\/_span> and webmap<\/_span> which includes the nationwide dataset Woningtypering. This data will be updated monthly.<\/_span>
If you have any questions, comments or ideas for applications of this dataset? Then we would love to hear from you! My colleagues and I can be reached via content@esri.nl<\/_span>, or of course just via a comment below.<\/_span>
Interessante analyse en een nuttige dataset.
Onze vraag is; als gevolg van deze analyse worden geschakelde 2-onder-1-kap woningen getypeerd als 'tussenwoningen', een typering die niet het juiste beeld geeft. Wordt er nog gewerkt aan een verfijning van de analyse zoals die nu is om een woningtypering '2-onder-1-kap - geschakeld' toe te voegen, of eventueel deze woningen te typeren als 2-onder-1-kap.
Geschakelde 2-onder-1 kap zijn over het algemeen dmv een garage geschakeld, misschien een uitgangspunt voor een verfijning van jullie analyse?
Beste Bert Jan, bedankt voor je input! Helaas is een dergelijke analyse op dit moment niet mogelijk. Voor dit soort woningen wordt in de BAG de garage niet apart ingetekend maar als onderdeel van de woning. Dit betekent dat het in een analyse niet mogelijk is om een aangebouwde garage apart te zien van het woonhuis en deze woningen dus niet te onderscheiden zijn van tussenwoningen. Vandaar dat deze woningen in de categorie "tussenwoning/geschakeld" vallen.
Signed in members can post, follow updates, and more. New here? Register a free account.
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.
