Logistic Regression (page 3 of 6)

The logistic regression algorithm attempts to minimize the losses across all of the observations. The losses are the actual value minus the predicted value for a particular observation. A model's total cost is the sum of the losses across all observations.

In this animated gif, an "epoch" is an entire transit of the training data through the algorithm. The epoch hyperparameter (not available in all algorithms in sklearn) specifies the number of full passes (epochs) of the entire training dataset through the algorithm’s training or learning process.