Can someone please explain what images of these kinds (attached below) are supposed to represent? I have seen them many times both in courses 1 and 2 but can’t understand them exactly. Sorry if it was clarified somewhere in the course. I probably missed it.

That is showing the ‘decision boundary’ of the model. In that particular case the input samples are points in the 2D plane which are either red or blue. The model’s goal is to predict correctly the color of the point based on the coordinates. The blue area shows where it will predict blue and so forth. The white line is the decision boundary. You can see in that particular case that the predictions are pretty accurate. In the case of logistic regression, the decision boudary is always a straight line in 2d or a hyperplane in higher dimensions. A neural network can produce a decision boundary that is non-linear and quite complex, which shows that NNs are more powerful than LR at this type of classification problem.

1 Like