CODEWITHSUNDEEP
machine-learningevaluation
bolleke
bolleke

Reputation: 151

Why is it important to have more than one evaluation method for different machine learning algorithms?

Couldn't find a precise and concise answer. I'm not particularly interested in different machine learning evaluation methods, I just want to know why it's important to have more than one?

Upvotes: 0

Views: 35

Answers (1)

avvinci
avvinci

Reputation: 386

Each metrics gives a different insight and evaluates your model differently. Let's take an example for binary classification:

  • Accuracy tells you what percentage of your predictions are correct. But what if you also want to know exactly how many 1's you got wrong [i.e. you predicted 0's where they should be 1]. for this, you will calculate the recall score.

So you get the idea maybe you want good accuracy but also good recall [real world example : maybe spam detection], so you look at both metric and choose wisely

Upvotes: 1

Related Questions