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

Polynomial Regression. Our next type of regression is called a polynomial
regression. In the last two types of regression, our models created a straight
line. This is because the relationship between our X and Y are linear,
meaning that the effect X has on Y doesn’t change as the value of X
changes. In polynomial regressions, our model will result in a line that has a
curve.
If we tried to use linear regression to fit a graph that has nonlinear
characteristics, we would do a poor job of creating the best fit line. Take the
graph on the left, for example; the scatter plot has an upward trend like
before, but with a curve. In this case, a straight line doesn’t work. Instead,
with a polynomial regression, we will create a line with a curve to match
the curve in our data, like the graph on the right 

---

The equation of a polynomial will look like the linear equation, with the
difference being that there will be some polynomial expression attached to
one or more of our X values. For example:
Y = mX2+b
The effect that X has on Y changes exponentially as the value for X
changes.
Support Vector Regression. This is another important tool for data
scientists and one that you should familiarize yourself with. It is most
commonly used in case classification. The idea here is to find a line through
a space that separates data points into different classes. Support Vector
Regression is another type of supervised learning. Its also used for
regression analysis. It is a type of binary classification technique not related
to probability.
In support of Vector Regression, all your training data falls into one
category or the other. You want to find out which category a new data point

---

falls into. Your data is separated into these two classes by a hyperplane.
When you're creating a model for the hyperplane, you are trying to find a
hyperplane that maximizes the distance between the two classes. For
example, in the next picture, you have a scatterplot where the data points
can be separated into two distinct classes. In this instance, lines one and
three can separate the data points into two distinct classes. However, for
your model, you should choose line two because it maximizes the margin
between the two classes, so they are more distinct. The wider the margin is,
the better.