Alteryx apps for the IoT wave?

Question

Well I couldn't be able to find much about Alteryx and "Iot" best practices in the same page...

I'm sure about the "Alteryx Designer"s data pre-processing and modeling capabilities are sufficent for IoT analytics.

But how about running the workflow for the Iot?

1) Is the analytic app will be on the cloud, "the gallery" and the Iot device will upload the data in realtime and receive answers as API result in JSON or

2) We will upload the analytics app onto something similar to an arduino board and run it, streaiming sensory data will be pre-processed, learned and acted upon on the go and on the tiny microcontroller?

I'm dreaming of something similar to a Nest termostat but the self learning and decision model will be prepared thru Alteryx and you can update the logic anytime you want, thru a bluetooth connection or towards the IoT devices own microsite, what do you think?

Best

P.s you can run java and with some effort Python on top of that miniture device...

chris_love · Accepted Answer

I think this is an interesting use case around the Internet of Things but there are going to be two problems as far as I can see.

1. Alteryx is not "real time", while we could schedule Apps to run via the scheduler every minute say we wouldn't be able to injest data in true real time. A much better option woudl be to have the IoT device write direct to a cloud database and run Alteryx models on that incrementally to create actions.

2. Having Alteryx run directly on a device is a very long way off - I think there is more interest in a Mac version for now :-)

There's certainly a great use case for Alteryx around the IOT, PMML support will certainly help that process in the long run, however for now the clear benefit of Alteryx is in quickly developing and testing these modules, deployment can then be moved outside Alteryx dependent on the device / datasource.

Atabarezz · Accepted Answer

Totally agree that it's impossible to load alteryx on top of that tiny device with just 256 mb ram etc.

But running the decision tree requires if then else statemens fairly short in terms of coding or regression equations requires simple java codes similarly...

My suggestion is converting the app into java or python or r which can be directly called by java... and loading that on top of arduino or raspberry pi. One of my suggestions was related to this; http://community.alteryx.com/t5/Alteryx-Product-Ideas/Alteryx-workflows-to-run-in-Java/idc-p/4098#M1315