In recent years Machine Learning has become a favourite tool for pretty much everything and everyone but, why is that? Is it really that good or is it just like a new song by Lady Gaga? (popular until a new song from Shakira comes out).
What is Machine Learning?
Machine Learning is a field of Artificial Intelligence based on statistics and probability to determine patterns in data. The interesting part is that the algorithm will determine how important are certain fields on that data, making the process more “pure” as everything will just fit in its place.
Take this common example:
If you had data like the one on the chart you could probably think that the number of bedrooms and Square footage are the most important characteristics in determining the price of a house.
With this small table with just 5 fields it’s easy for a human to think about a price for a different house but, how accurate is that number? Also, in the real world you would have a ton of other fields like the country, the state, the area where the house is, if it’s winter or spring, if a renown architect designed it, how old it is, etc.
The complexity doesn’t end there, because you’ve got these many fields you would also need a lot of examples to get an accurate idea of what is actually happening, and yeah, it’s a very big number of examples! This is where Machine Learning helps.
Machine Learning algorithms “study” each one of the examples and try to understand how they determine the price of a house. With each new example the algorithm refines a function that will approximate the price later on. At the end you might get very useful and accurate data.
Here you can clearly see the many uses of Machine Learning, maybe to understand how someone uses a device and therefore when to update apps, or maybe to try to give him a better experience in using one.
Machine Learning has become popular due to many factors:
We produce large volumes of data
Mining isn’t enough to learn from what we produce
Data is now the most valuable resource
Governments, enterprises, politicians, all of them would like to learn from us: what do we like? how much? Why? These are questions Machine Learning algorithms can help us solve.
