My name is Jonathan.T, welcome to my page. I enjoy solving problems using data, math and statistics. I am currently working in the analytics team as a senior data lead at SCOR based in Sydney, Australia. I am also the co-founder & CTO of the data consulting company, Bastion Labs.
Feature Engineering Explained
What is feature engineering and why are we taught to add seemingly useless
variations of
certain features
to our dataset in the hopes of a "better prediction"?
The plot shows a very simple example of a 9-point dataset. Say we wanted to
model a
straight line to separate/ classify the red and blue points.
Without bending the line, this would be impossible. What happens when we add an
extra
calculated column to the dataset?
Click on the "Add feature" button to find out!
Now see if you can then use a flat plane (the straight line equivalent in 3d)
to split the red and blue points.
Simple Linear Regression
The Simple Linear Regressor is the OG machine learning algorithm and is
constrained
by a
line. It uses the Least Squares method in order to calculated the "best-fit
line" in
a
given dataset. Try out the widget
to see how close
you can get to the best fit line by manually tweaking the parameters and read
the
full
article to learn more about Simple Linear Regression.
Parameter settings:
Errors:
Minimized Error: