New

Time estimation for learning process ml

  • 11 September 2020
  • 2 replies
  • 33 views

Badge

I am currently testing ml on my laptop. So I have sqlserver, indicium and ml-docker image running locally.
Training a model takes quite a long time. In my first tests I tried to train a table with 100,000 rows with 4 predictors. I left my laptop on and 8 hours later it was still training the first model.  
It would be useful to get a time estimation of how long it will take to train a model, so that it is clear whether waiting for the process is useful or whether it is better to use another dataset.
 


2 replies

Userlevel 3
Badge +1

Here, have my vote!

 

Userlevel 3
Badge

We already have this wish on our backlog, meaning that we will start with developing this feature somewhere in the near future. Besides that we are also investigating means to speed up the process, especially for larger data sets.

 

Reply