Scala and Spark for Big Data Analytics

Binary classification using logistic regression

Logistic regression is widely used to predict a binary response. This is a linear method that can be written mathematically as follows:

In the preceding equation, L(w; x, y) is the loss function is called logistic loss.

For binary classification problems, the algorithm will output a binary logistic regression model. Given a new data point, denoted by x, the model makes predictions by applying the logistic function:

Where z = w^Tx, and by default, if f(w^Tx)>0.5, the outcome is positive, or negative ...