Logistic Regression Machine Learning Algorithm

As explained earlier, Machine learning is divided into two main Branches. Supervised Learning and Unsupervised learning.

The
Supervised Learning is further divided into two types of learning Models,
Classification learning Model and Regression Learning Model.

Unsupervised Learning is further divided into Clustering learning Model and Association learning Model.

When do you use Supervised Learning Model

Supervised learning Model is used when the available data is labeled.

What
are some Examples of labeled data?

Just imagine you have a set of patients coming in to your clinic. You wish to Classify them on the label, “Healthy” and “Unhealthy”. Here, the labels used are Healthy and Unhealthy.

Based on the information available you can classify any unlabeled data that comes to your clinic. The Table below gives an Example of Unlabeled Data.

Types of Classification

There are many available types of Classification too, example, Binary classification, Gender classification, File type classification, Size type classification, Classifying data on security types, like Private, Public, Restricted.

Classifying data can also be applied at multiple levels, example given 1000 people, you will be able to apply classification process to classify, based on body types, skin color, hair type, etc.

Some Note worthy Points

Classification learning Model is always applied to data that
is discrete in nature.

On the other hand, Regression leaning Model is applied when data is continuous in nature. An Example for Regression Model would be, given 1000 students work ethic and related test scores, you would be able to predict the next 1000 students test scores with the Regression model designed.

The Logistic Regression Model

In Logistic Regression model, the outcome is always straightforward. When we say straightforward, we mean, when the data is input, the decision is got. For example, you wish to classify whether a person is Fat or Fit, whether a person is Tall or Short, Healthy or Unhealthy, Good Chef or Bad Chef, Good Student or Bad Student, Good Teacher or Bad Teacher. The decision is instantaneously achieved from the Logistic Regression Model.

Difference between Linear Regression Model and Logistic Regression Model

In Linear Regression model, we fit the Line to the Data, but
in Logistic Regression Model, we fit a “S” shaped logistic function to the
data.

The curve gives the information on the probability of a person being Fat or a Fit based on having a Gym membership.

If the person is Heavy, then there is a high Probability that the person will not hold a Gym Membership.

On the other hand, if we are to weigh in a person who is
between Heavy and moderately Heavy, then the Probability of the person having a
Gym Membership will only be 50%.

Threshold Value

0.5, is taken as the threshold value. If person has values
lesser than 0.5, then he or she is classified as Fat. If the person has greater
than 0.5, he or she is classified as Fit.

In the model, Fat or Fit, is predicted by Gym Membership.

Important Note

Please be informed, this is merely an illustration, of Course we understand, just by being a member in a Gym, does not automatically make a person Fit, on the contrary, the Model assumes that all who have Gym Membership will hit the Gym for working out purposes only and not for other curricular interests.

Logistic Regression Models can work with both Continuous Data set as well as Discrete Data set.