Tips & Tricks: Exporting Scalable Images from RapidMiner Studio
One of the most appealing features of RapidMiner is the visual representation of the analytical processes you build. Not only does it make designing a data science work flow an intuitive experience, it also makes it very understandable for others. So this post is not about presenting your results in a polished way, but about how you can do the same with the underlying RapidMiner process and export scalable images.
Capturing your ideas in a meaningful picture is a powerful way to communicate those ideas. This is why capturing a screen shot of a RapidMiner process always is a good way to show your ideas either to your boss or a customer.
A simple screen shot might be sufficient for your typical business meeting slide show or web session. But it isn’t adequate if embedded on a high resolution web page or in print, e.g., for a news article, a research paper or a poster presentation. On a large conference screen or A1 poster, a blurry, pixelated picture just doesn’t look decent. No matter how good your ideas are, the eyes are always drawn to the pictures first and first impressions matter.
Don’t despair! I’m here to share an elegant solution. RapidMiner Studio offers the option to export a process in the Scalable Vector Graphics (svg) or Portable Document Format (pdf) file formats. As the names suggest, images in these formats are scalable and don’t suffer quality loss when enlarged, in contrast to your typical jpg or png.
Exporting a RapidMiner process as a svg file
So now we have the process in a nice format that scales well, but we can do even more. With the right tools for image processing (e.g. the free to use Inkscape) you can enhance exported process even further.
Nearly every element in the image can be manipulated on its own. To do this you have to ungroup all elements, as shown below. Then it’s possible to rearrange, manipulate or delete specific elements of the process.
Single components of an Operator
For example, I start by cleaning up a bit and remove unnecessary extra icons, that might distract from the main process. Also slightly bigger process icons and names look good. Tip: when selecting all operators (by hitting Strg+A) cut off long operator names are extended.
Simple screen shot of a process
Improved process after editing the svg file
Depending what and where you present the image there is no limit for your creativity. Here are two examples from research posters which include a RapidMiner process.
For presenting the PRESED research project at the European Data Forum 2016 the goal was to highlight the process workflow and the different technologies involved. The process is cleaned up and the icons are slightly enlarged, as described earlier:
The second poster was for a demonstration showing the upcoming HypGraph developed for the FEE research project. The new operator is enlarged as a whole, to highlight its importance. I also used the fact that there is a white halo around the operators and connections. So even with the dark background everything is still readable.
The final posters are linked below, so you can see how the images are embedded in the overall presentation.