Intuition: Training Set vs. Test Set vs. Validation Set

The difference between training, test, and validation datasets can be confusing. Especially when you think that testing means to create tests and compare different results between models and validation means to verify those results. That's what I originally thought when I learned it... apparently that's wrong.
More videos: / intuitiveml
Follow:
Twitter: / sentimofficial
Facebook: / sentimofficial
____________________
Learn more:
Details: en.wikipedia.o...
Less details: machinelearnin...
Even less details: stats.stackexc...
#IntuitiveML, #MachineLearning, #AI

[@jamesang7861]

I always remember it this way, to simplify my own understanding.
Training set: The effort we put we're studying for an examination (training our brain).
Validation set: The sample exam test papers we do to test our understanding on what we have learnt.
Test set: The actual final exam that we take.
Or course, the 'test set' could also mean real world application of our knowledge/understanding, depending on how we want to put it.

[@IntuitiveML]

Well said!

[@VFORVICTORY1000]

thx it helps me a lot

[@priyalbhardwaj696]

Woah it helped me thnakss ☺️☺️☺️

[@r0cketRacoon]

well said, short enough to enlighten me. Tks bro

[@utkarshverma3916]

Short but crisp and effective, crystal clear👍👍

[@IntuitiveML]

Apologies! I thought I published this a month ago, but I actually forgot to move it from unlisted to publish until now! That's my bad...

[@sujayghosh3540]

Not a problem

[@purple3036]

Wonderful explanation, thanks for the video

[@mikemihay]

Great video! Thank you for creating it, good visualisations

[@Yassinius]

Great video!

[@andreamansi2441]

A nice refresh :)

[@IntuitiveML]

I'm glad it was helpful for you!

[@biruhandualem9930]

how can I communicate with u individually? like telegram account............

[@daifang2029]

The most clear explanation I have ever seen.

[@barax9462]

So basically after training the weights,, I just should pass the validation into forward-probagation and compute the loss??

[@IntuitiveML]

Normally you compute the metric you care about with the validation and test sets, such as accuracy, precision, and recall

[@neelshah1651]

The most clear explanation I have ever seen, Keep it up...

[@잠자기장인-x3r]

Thank you a great video that helped me understand in barley over 2 minutes

[@lionghanwen257]

how come the others "expert" cannot explain this in such a simple and easy understand way? You deserve my 5 stars

[@darwin6984]

Most ppl combine validation and test set.
May you kindly advise How to determine how much percent for 3 datasets? and methodology to determine it?

[@mariap.9768]

Perfectly concise.

[@DoubleMagnet]

This video is very helpful, thank you so much

[@luzzyommm]

this is the best explanation of validation vs test data!

[@NosurfOfficial]

Thank you, please keep making more videos, this was really helpful.

[@IntuitiveML]

More to come very soon! I'm making a longer video and I'm a little behind...

[@md.nobihossen6183]

thanks a lot. my confusion is cleared now after watching your video. best of luck for the future.

[@IntuitiveML]

You are most welcome!

[@li-pingho1441]

this is amazing

[@lucarauchenberger628]

gold! great explanation!!

[@IntuitiveML]

:)