From the course: Machine Learning and AI Foundations: Classification Modeling

Unlock the full course today

Join today to access over 22,600 courses taught by industry experts or purchase this course individually.

Confusion matrix

Confusion matrix

From the course: Machine Learning and AI Foundations: Classification Modeling

Start my 1-month free trial

Confusion matrix

“

- [Instructor] Okay, let's talk about he confusion matrix. Now folks will joke about the name but, the whole idea behind the name confusion matrix is where the model is making errors. And it's one of the most fundamental techniques we have for assessing our model performance. So let's take a look at the first thing you should probably look at, which is overall performance on the train versus overall performance on the test. Here, overall performance on the train is 79.8%. Overall performance on the test data is 74.2. That's a gap of a bit more than 5% which is just a hair outside the comfort zone. Meaning that, we fear that the model isn't all that stable. It's performance on unseen data just isn't as strong as the performance on the data that we fed to the modeling algorithm. It's a close call in this case but it's at that borderline and it's kind of on the wrong side of the border. We want it to be within 5%. So let's now talk about the wealth of data that's hiding in here but…

Contents