In: Computer Science
Please describe Logistic Regression classification methods and its usage?
Logistic Regression:
Logistic regression is similar to Linear regression other than the
fact that, logistic regression classifies something into true or
false, instead of predicting something continuous like something
like size of a person, or height of a person.
Instead of fitting a line, Logistic regression fits an 'S' shaped curve called the sigmoid function.
Logistic regression can be called as probabilistic classification of a data that means what is the probability of an animal being a dog or a cat based on the data like tail length, height, weight etc. of the animal.
With Linear regression we fit the line using "least squares"or we find a line that minimizes the sum of these residuals.
Logistic regression uses maximum likelihood. In a nutshell, you pick a probability, scaled by a factor that affects the classification. Like in this case, obesity of a mouse.
So to summarise, logistic regression is used to classify data on the basis of probability of a class being what it is. If the probability extracted by processing the factors is inclined more towards a particular class, then the example is classified in it else, the other class.
Basically Logistic regression is used to classify categorical data like cat or dog, boy or girl on the basis of features, play or not play on the basis of weather conditions etc.