Engineering Sandbox rollout ยท bundle-mounted ML essays

Move the threshold, then see probability turn into classification.

Logistic regression is easier to trust once you can watch the sigmoid, the boundary, and the classification decision shift together. This route pairs that intuition with loss and coefficient scenes so the model stays connected to its mechanics.

Move the decision boundary Read log loss Watch coefficient updates
How to use this page

Start with the temperature and threshold controls until the prediction flips in a way that feels intuitive. Then scroll into log loss and gradient descent so the probability story connects cleanly to training.

Play first: move the boundary Jump to evaluation