Introduction to Machine Learning (10): Regression and Clustering Algorithms-Logistic Regression and Binary Classification

Learning Directory:

Logistic regression content directory:

Logistic regression application scenario : applied to the binary classification problem.
Logistic regression principle : map the output of linear regression to the activation function sigmiod, and output a number in the 0-1 interval as a probability value. If it is greater than the threshold we set, it is considered to belong to This category.

Loss function: log likelihood function

The overall logistic regression process:

API：

Classification evaluation index

Calculation of precision rate and recall rate:

When the sample classification is not balanced, 99 no, 1 yes, it is not easy to use the precision rate and the recall rate : it is necessary to introduce the ROC curve and the AUC indicator



API

Model saving and loading