At first MLA+ invited me without a fixed goal in mind. They knew their need to research urban data to get location-based results and I was the guy to come to with this kind of request. Together we had a couple of meetings, where we pointed out objectives and our plan to meet them. After that, I've put together a list of service types to download and approved it with MLA+ guys. Than I've downloaded everything with the help of Python (now my dear work friend), put all data into one file and threw it into QGIS.
After a few experiments on the size on hexbins for our research, we've stopped with 65x65 meters for understanding city center better — this size perfectly corelating with typical central block, and 150x150 meters for city-wide research and corelation with all other city blocks.
Having done the hexbin map, sanity check was in order, to make sure the data and research was going okay. There were a couple of topics I was curious about during the check, one of them being highly dense hexbins far from metro stations and far from city center. Looking at raw data in the form of points, I've found out that those spikes of density were hospitals, each of the departmnets showing as an individual point. Although I had a temptation to even this out by leaving one point for each hospital, at some point it came to me that hospitals are, in fact, local centers for many people, but common perception is different because of lack of other types of services and urban environment around.
Half a year after finishing the project, I've learned the art of making 3D data maps in Mapbox and couldn't help myself making this map in 3D. To top it off, I've done my own deeper research on services data in a form of
Medium post — looking at various corelations like densiy of services and density of built environment.