The magic of features: Transforming datasets

What do you do when given a dataset which is not linearly separable?

Answer: Take data from one dimension and transform it into a higher dimension to make it separable.?

We now take feature representation for granted, but it is such an awesome idea. However, you need to see it visually to truly appreciate it.

In Day 13 of the ML: Teach by Doing Project, I dive into feature representation and data transformation.?

Here is how the original XOR dataset looks like in 2 dimensions:

It is not linearly separable.

Now let’s do the feature trick. We will transform the 2D data into 3D.?

This is how it looks like now:

We clearly have a linearly separator in 3D! Isn’t this amazing??That's the magic of features and how they can help transform data.

Here’s the video of all my learnings regarding the feature representation trick:

My lecture notes and code files have been added here: Link

Stay tuned for Day 14!