I'm trying to calculate distances between geographic locations given as (longitude, latitude) pairs. Both values are given as Double in a CSV-File.
I build a KNIME-workflow as follows:
CSV-Read |--> Row Sampling | --> Distance Matrix Calculate
Unfortunately, when I attempt to configure the [Distance Matrix Calculate] Node with the Haversine Distance, I receive a "no column selected" error message. Conversely, when I select the Euclidean Distance the columns remain selectable. Can anybody tell me why the columns disappear/do not provide the required data format for the Haversine Distance?
I suppose what you see is the Haversine distance implementation from the Palladian nodes (if you have not installed them, another Haversine distance might have been added to KNIME of which I'm maybe not aware of :) ). This distance measure requires the column type "GeoCoordinate".
Unfortunately, we had to remove the geograhpical nodes from the community nodes distribution of Palladian, because we cannot provide their functionality completely for free (this is due to the fact, that our LocationExtractor node requires API access to our server to work, and we cannot provide unlimited API invocations for free). However, we're working on a separate Palladian release, which will include all the geograhpical nodes (MapViewer, LocationExtractor, CoordinateParser, distance measures, aggregation functions, etc.). I hope, that this release will be ready within ~ 4 weeks, but I can give no definite promise.
So, in short, the problem is currently, that the distance measure shows up, but the nodes which are necessary for creating the required data type are not available in the Palladian community distribution.
If you need this functionality urgently, contact me and I'll see if I can provide you with a custom build in the meantime.