Microsoft's Drag and Drop machine learning tool

Microsoft's Drag and Drop machine learning tool

by Daniel Neis Araujo -
Number of replies: 2
Picture of Core developers Picture of Plugin developers Picture of Translators

Hi,


as I didn't find a better place to post, here is the link:

https://techcrunch.com/2019/05/02/microsoft-launches-a-drag-and-drop-machine-learning-tool-and-hosted-jupyter-notebooks/

"The new interface for Azure’s automated machine learning tool makes creating a model as easy as importing a data set and then telling the service which value to predict. Users don’t need to write a single line of code, while in the backend, this updated version now supports a number of new algorithms and optimizations that should result in more accurate models. While most of this is automated, Microsoft stresses that the service provides “complete transparency into algorithms, so developers and data scientists can manually override and control the process.”

It would be real awesome if we can have anything similar to that to create models on Moodle.

Average of ratings: -
In reply to Daniel Neis Araujo

Re: Microsoft's Drag and Drop machine learning tool

by David Monllaó -
Thanks for sharing the link Daniel.

Yep, it would be great to have an ML backend using this service internally. Google (autoML) and other companies offer similar services. The problem is that most of those services, including the one you are sharing (please correct me if I am wrong), require payment and are not open source. We can not compete with them in terms of features but given Moodle's open source nature it may make more sense to continue pushing our efforts towards free and open sourced solutions. If any 3rd party is interested on integrating any of these services we can support them updating our APIs to make their life easier. If they share the code the benefit is for everybody.
In reply to Daniel Neis Araujo

Re:

by Elizabeth Dalton -
If I understand you correctly, you are not suggesting that we create an integration with the Azure tool, but that we have functionality that would make it easier to create new models, like the Azure tool. This tool really isn't quite as simple as they are implying. The data set you import consists of pre-calculated features and a pre-calculated target-- it's not a generic data set like Moodle logs. ;) The functionality they are showing is really not much more than we have in Moodle 3.7, where you can define a new model and select a target and indicators (features) from those available. There are some additional algorithms Azure offers that we don't have yet in Moodle, but we're working on those. (See our Google Summer of Code student projects!)

One thing I like, however, is the "data profile" view, which shows all the features in your data set with some statistics about each (including a histogram). I think we could make a report like that-- even in Configurable Reports. smile