Grouping Analysis with streets as borders?

[IsabellRuehl]

Hello,

I have a question about the Grouping Analysis Tool.

Is it possible to use streets as spatial borders for the groups?

I want to group a city district with multiple attributes, like the heat demand of buildings and the number of building types. But spatial neighborhood is of primary concern.  I have a data set that contains 149 blocks of buildings in the form of separate polygons. And I have another one which contains streets. There are two main streets and I don´t want the created groups to cross them in the sense that there are parts of one group on both sides of a street. Is there a possible solution?

Or is there another tool or script which I could use in ArcGIS to create spatial clusters in the presence of obstacles? Unfortunately I couldn’t find anything related to this using the search function.

Thank you all!

[SepheFox]

What about treating the buildings as separate data sets, by manually selecting all the buildings in each separate area? Since there are only two streets.

[SepheFox]

I think you could also add an analysis field, that you calculate to be a unique integer for each "group" of buildings between separating streets.

[IsabellRuehl]

Thanks a lot for your help!

I added a new analysis field and set different integer values (1, 2, 3, 4) for each separated area. In that case there were four separated areas. As you can see in the attached image the result was quite good. But the green group is still crossed by a street. Do you think this is due to ArcGIS’ priorisation of spatial vicinity when selecting spatial constraints?

Do you have maybe another solution? Or do you know another tool or something?

The problem with manual selection is that this procedure should be applicable to different city districts with multiple attributes.