New exported format [offline data exploration]: Orange Data Mining project file (generic .ows)

[fititnt]

• Software
  • https://en.wikipedia.org/wiki/Orange_(software)
  • https://orangedatamining.com/
    • https://orangedatamining.com/download/
    • Some tutorials
      • https://orangedatamining.com/blog/2020/2020-04-02-covid-19-basic/
• Theory and concepts
  • https://en.wikipedia.org/wiki/Statistics
    • https://en.wikipedia.org/wiki/Data_mining
    • https://en.wikipedia.org/wiki/Machine_learning
  • https://en.wikipedia.org/wiki/Decision_support_system
    • https://en.wikipedia.org/wiki/Recommender_system
• Related
  • https://github.com/EticaAI/HXL-Data-Science-file-formats
  • https://github.com/EticaAI/HXL-Data-Science-file-formats/issues?q=orange
    • hxl2tab: tab format, focused for compatibility with Orange Data Mining HXL-Data-Science-file-formats#2

---

From the entire EticaAI/HXL-Data-Science-file-formats project, the Orange Data Mining was one of the easier point-and-clicky (even among the ones with interfaces) to do statistical analysis and machine learning based on data. It still takes more memory ram than Weka (or even low level with python) but still reasonable considering most audiences would use Excel for it.

The idea here is to make proof of concepts of generating orange project files with some basic functionality to analyse data.

Know limitation

Even if we manage to make some templated orange project files (the way we do now with datapackage.json and csv-metadata.json) orange can even try to infer the categories of the data, but it will fail to decide which is "target" which the user wants to discover the relations.

Even for datasets which might have data that would worth analyse with Orange, we may either put some dummy target or leave to user select. In this sense, it not strictly an limitation, but we can't automate this part.