Author Topic: 15 Beta Data Miner Feature expieriments  (Read 1751 times)

Ian Mitchell

  • EA User
  • **
  • Posts: 344
  • Karma: +15/-4
  • The eaDocX and Model Expert guy
    • View Profile
Re: 15 Beta Data Miner Feature expieriments
« Reply #15 on: July 16, 2019, 05:47:02 pm »
This is a great piece of work, Guillaume  - well done! An excellent guide to the mechanics of making the Data Miner work.

Sadly, this looks like another example of Sparx providing us with, say, and engine, a bunch of metal bits and welding torch, when what we want is a way to drive to work. It's kind-of necessary, but very far away from a workable transportation solution.

My set of requirements (below) still stand as a minimum set for a workable "ETL" suite. In particular, we're still missing tools to:

1 - analyze examples of proposed input data, to describe what they contain. EA should be brilliant at this bit.

2 - map each row/cell of the input to the output, using a GUI, and not a load of code. After all, most of these scripts which I need to write will say the same thing - take attribute A from file B and make it attribute X of element Y. Attribute-level mapping - hey, we can do that too, especially with V15.

3 - manage this, so as to make it reliable and auditable: where's the boring stuff like file management ("did the right file arrive in the right place last night") and management of updates ("Last night's File A says that element Y needs to be updated - who should I tell?"). Ok, not so easy, but Prolaborate might be able to help here?

I wonder if the assumption @Sparx is, yet again, that all EA users are programmers.

Hot news - they aren't.

And if you make them write code to make stuff work, then they won't, and they will buy other tools instead.

@Sparx - please spend a few minutes googling '"ETL tools" to see where the state-of-the-art is right now: this looks like a solution which someone might have cooked-up 25 years ago, who hadn't heard of ETL.
Ian Mitchell, Designer, eaDocX