Can I do Python Machine Learning (using keras library) in TagoIO?

I have installed several sensors which they send their data to TagoIO. Here I wan to use a python code to use the stored data and do predictions based on Machine learning. This process repeats one each day, so the python script should get all of the stored data, train a model, and use it to predict the next day.

  1. Can I use Python, together with keras and other required libraries for data mining here?

  2. Training of model may take several hours, is it ok in TagoIO or there is a time limitation?

  3. Considering that one run of my Machine Learning script may take several hours, is it calculated as 1 run in cost calculation?

  4. Apparently there is a possibility to run external programs. Can it be a solution?

I am very new to TagoIO and also to IoT so I really appreciate if the answers are simple to understand :slight_smile:

Hi @amirreza.heidari,

You not able to add 3-party libs on python analysis for now, (you can do that on node analysis using tago-builder)

Analysis on TagoIO has a timeout of 30 seconds.

You can’t run an analysis on TagoIO for more than 30 seconds, but yes every time you run an analysis it will cost 1.

Run analysis external is a great solution for that problem, you can run analysis in any language and runtimes, also you can access local database (on the intranet) the connection with TagoIO is pretty simple it uses a socket connection, it’s mean you don’t need to take care about open ports or nothing like that, and it’s for free btw.