▶Probability model
The conditional probability is the sigmoid of a weighted feature score.
Logistic Regression
Probabilistic classification for binary outcomes
Logistic regression sends a linear score through a sigmoid so that the output lies between zero and one and can be interpreted as a class probability.
watchone idea
→manipulateone example
→leave withone intuition
The essential idea: a linear decision boundary can produce calibrated probabilities through the sigmoid link.
One Concept · One Example
What to notice
- Weights and bias form a linear score.
- The sigmoid maps every score into zero to one.
- A threshold converts probability into a label.
A line becomes a probability
Training loss
Binary cross-entropy rewards confident correct predictions and strongly penalizes confident mistakes.
Move the score and the classification threshold
Adjust the weight, bias, and threshold. See the probability curve and resulting decision change.
Logistic regression couples a linear score with the sigmoid to produce trainable binary class probabilities.