Timeseries in Workspace

Ask the Community
, ,
1

I am trying to use multiple timeseries to create an input dataset for my model.
I have a couples of timeseries of the raw signal, and another timeseries that determines when my signal has completed a cycle.
These raw signals then can go into my autoencoder and I can determine if the signal is normal or not.

What is the best way to tackle this challenge and is there a tutorial out there for the same? I code in python. I find the documentation quite sparse.

I’m guessing code workspaces is the best place to do this.

2

Hey @rho,

Interesting question!

Just to clarify a few things for your setup:

  1. Where is your autoencoder model hosted?
  2. How up to date does the encoding need to be? Does it need to be live or can it be done in pipeline?
3
  1. It is not hosted yet. I plan to deploy it within Foundry itself. I’m having some struggles there too. For the purposes of this question, I would say it is flexible. I want to take the path of least resistance. I thought about having the data transformation happening in the model adapter - maybe that is where it should be done? I’m not sure.
  2. I would love to run live, but as I said even getting something out there would be a success. It can be a pipeline. Which is also why I started with Code Workspaces.

PS. Appreciate your help, and prompt reply.