Data mining workshop in Kuching

Earlier this year, my boss (Geoff Holmes), gave an invited talk at the CITA’15 conference in Kuching, Malaysia at UNIMAS/FCSIT. During his talk, he mentioned ADAMS being used for (kinda) big data in plant/soil analysis using spectral data, allowing for large savings (for one of our customers these are between 18 and 33 million USD per year). This attracted their interest and turned into an invitation to give a two-day workshop on ADAMS in Kuching.

The workshop took place last week on November 26/27 and it has been fairly successful, I have to say. People there (staff and students) were familiar with workflow systems, having worked with systems like Kepler and RapidMiner. ADAMS is a bit different, by having no explicit connections and using a tree layout. However, the attendants also attested that using the (initially) more intuitive, canvas-based approach that RapidMiner and similar systems use, became quite tedious for larger workflows. With ADAMS they were able to keep a better overview of things, simply by collapsing sub-trees in the workflow. They have plans now on using ADAMS for some of their projects and, with a little bit of help, they should be able to reach that goal. 😉

Anyway, if you are interested in the workshop material, you can grab it from my github repository:
https://github.com/fracpete/kuching-2015
The material is licensed under CC-BY-SA 4.0 and GPLv3. However, you no longer have to use a custom build of ADAMS, as the incubating NLP module has been folded into the adams-addons group of modules. You can simply download an adams-addons snapshot to run the example flows.