LFD Book Forum  

Go Back   LFD Book Forum > Book Feedback - Learning From Data > Chapter 4 - Overfitting

Reply
 
Thread Tools Display Modes
  #1  
Old 08-10-2021, 12:51 PM
gverhoev gverhoev is offline
Junior Member
 
Join Date: Jul 2021
Posts: 6
Default Does a test set exist?

In his recent lecture on testing and validation, Professor Magdon-Ismail said that a test said is used in a fictitious world. Does that mean that the term/concept "test set" does in reality not exist, as one always splits a data set in training and validation sets? And as soon as data are split, they influence the final hypothesis g because there are fewer data points to train. So can one even use the concept of a test set? Or is there a rule when one could use the term test set, also in a real-world situation? Many thanks for your insights!
Reply With Quote
  #2  
Old 08-11-2021, 04:36 AM
htlin's Avatar
htlin htlin is offline
NTU
 
Join Date: Aug 2009
Location: Taipei, Taiwan
Posts: 610
Default Re: Does a test set exist?

In my experience, a one-use-only test set usually does not exist in the real applications, given that "one-use-only" is arguably too luxurious in any application. When multiple tests are conducted on the same test set, the set becomes closer to a "validation" set than a test set. But anyway, IMHO those are just naming issues. The important thing is to know that every use on a set can cause some sort of "contamination."

Hope this helps.
__________________
When one teaches, two learn.
Reply With Quote
Reply

Tags
test set, testing

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 02:28 PM.


Powered by vBulletin® Version 3.8.3
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.
The contents of this forum are to be used ONLY by readers of the Learning From Data book by Yaser S. Abu-Mostafa, Malik Magdon-Ismail, and Hsuan-Tien Lin, and participants in the Learning From Data MOOC by Yaser S. Abu-Mostafa. No part of these contents is to be communicated or made accessible to ANY other person or entity.