Why Math?

So they said Linear Regression is easy and yes they were true. All that was needed was to make use of Python (I am being partial here you Excel romantics). Use a library that offers the functionality, learn what Y=mX+C means, draw a scatterd plot then find the best regression line. Voila!

Or say we're performing clustering using K-Nearest Neighbors (As the name suggests, i belong to the nearest neighbor when i don't belong anywhere else), the new data point will be classified to belong to a particular group which is the nearest to it. Sounds simple enough.

But if we look closely at the following figure, from the  theory stated above, all we did (or rather say Python did was) was find the distance of between the new point and all other points and classify the new data point to belong to the group which is the closest. In reality the theory is much more complex than the simple paragraph stated above. Lets keep the good stuff for later. Since I am NAIVE at this point of time, i would like to stick with the basic definition.

 

Now, Python is smart enough to get the job done for us which eventually results in getting a promotion or a performance award (well nobody wants to understand the HOW part, only the WHAT part). If we PAUSE for a moment and ask the IMPORTANT question, "HOW" did we calculate the distances, the experience would go beyond promotions and making false impressions. 

The consequent blog posts describe various aspects of Mathematics which form the basic building blocks of Machine Learning. The Geometrical intuition describing the complex concepts in art form makes the intricate theory more romantic (Being poetic here)

By the way the answer to the HOW is "Euclidean Distance". If that sounds intriguing, think of it as the "Length of Hypotenuse of a right triangle, by the good old Pythagoras".