Logistic Regression

YongBoYu 2024-05-14

• The logistic regression is a classic classification method. The motivation is to modify Logit(P(Y=1)) = β₀ + β₁X₁ + β₂X₂ + ... + βₖXₖ which transform the linear relationship of the feature into 0, and 1(binary outcome), we do this by using the sigmoid function. We can then calculate construct the probability function for each feature in the dataset.

• When we build the probability function. We build the maximum likelihood function that tells us by certain β. we will have the best result for this logistic regression.

• Then we optimize the MLE formula first by taking the log then use some optimation tool such as gradient descent or newton.