🎉 🎉. RAPIDMINER 9.8 IS OUT!!! 🎉 🎉
RapidMiner 9.8 continues to innovate in data science collaboration, connectivity and governance
Tutorial for the "JSON processing with jq" extension
The first step is to install the extension from the Marketplace if you don't have it yet. Click the Extensions/Marketplace menu entry, and in the Marketplace window, enter JSON in the search box. Select the extension for installation, and let Studio restart itself when the installation is done.
Let's play with a publicly available data set from the Vienna Open Data server.
Here's a list of playgrounds in the city of Vienna, with some attributes and even geocoordinates:
When you go there, you see a list of documents in different formats, we're obviously interested in the JSON document, with this URL.
The simplest RapidMiner process to get the contents of this URL and process them with jq looks like this:
In Open File, you set the resource type to URL, and paste the URL of the JSON resource. In Read Document uncheck "extract text only", as we don't want to change the input. Then add a Process Document with jq operator and connect its input and output ports.
By default, Process Document with jq is set up for JSON output, indenting (formatting) the resulting JSON document, and with the simplest jq expression ".", which just copies (and formats) the incoming document.
So we get the first result from the process:
The document contains a kind of a header (type: FeatureCollection and totalFeatures from the GeoJSON standard), and an array of "features" (the playgrounds).
We're interested in the name (ANL_NAME), the playground details (SPIELPLATZ_DETAIL), and the geocoordinates of every playground.
To develop the jq expression, we go to jqplay.org and paste the JSON data into JSON field. The we interactively begin to develop the expression to select the data we want.
The first step is ".features". This selects the features array (discarding the header), and returns every element as an object of one large array.
If we change this to ".features", we get a list of different objects, which is better for further processing.
In jq, we use the pipe symbol | for processing steps. Now we list the elements we want after a pipe. The expression is:
.features | [.properties.ANL_NAME, .properties.SPIELPLATZ_DETAIL, .geometry.coordinates, .geometry.coordinates ]
This gives us a nice flat structure that we can easily process with RapidMiner, especially if we let the operator convert it to CSV.
(Uncheck "first row as names" in the Read CSV operator, and change the separator to comma.)
The result is a normal RapidMiner example set:
Check out the documentation of jq if you need to process even more complex documents. jq offers additional functionality like extracting a variable length array of values (like tags) to a table structure, or counting elements, regular expression replacements, etc.