Machine Learning: 2 Books in 1: Machine Learning for Beginners, Machine Learning Mathematics. An Introduction Guide to Understand Data Science Through the Business Application

Classifications
A few times in this book, we’ve referred to classification models. While
some of the models we’ve already mentioned are capable of classification,
the following are more supervised learning models that are specifically used
for classification.
Classification requires labeled data and creates non-continuous
predications. In classification problems, the graphs are non-linear. There
could be two classes in a classification problem, or even more.
Classification models are probably the most widely used part of machine
learning and data science.
The first type of classification is binary classification. With binary
classification, the data is classified into two categories, denoted by 1 or 0.
We call it binary classification because there are only two possible
categories, and all of our data falls into one or the other.
But there are instances when we have more than two categories, and for
this, we use multi-class classification models. Also, we have linear decision
boundaries, where data is separated on either side of a line. Not all data can
be classified into either side of a decision boundary.
The first picture has an example of a classification using a linear decision
boundary. In the second image, there are almost two classes, but they are
not linearly separable. In the third image, data points are mixed, and linear
boundary classification is not possible. Depending on the type of data you

---

are using, there are different model choices that will be better suited for
different tasks
Logistic Regression/Classification
This method is used to classify dependent and categorical variables.
Logistic regression calculates probabilities based on independent variables.
It gives the variables the value “Yes or No” to sort them. Typically used
with binary classification.
When you can’t separate the data into classes by a linear boundary, like in
the examples above, this is the method you must use this. It’s one of the

---

most common types of machine learning algorithms. It doesn’t just sort into
categories, but it also tells us the probability that a category exists.
We denote this model by taking the odds function, where p is the
probability of an event;
And creating a formula called the logit