What you need to know about machine learning in the simplest terms possible. That’s it. 

Let me start off with this disclaimer: I am not a data scientist nor do I pretend to be one.

Now let me start off with this second disclaimer: You do not need to be a data scientist to understand the basic concepts of Machine Learning. Or Data Science. Or Artificial Intelligence.

Now I’m going to teach you a little about ML.

The cake

I like cake. So I’m going to request that you bake me a cake.

In order to bake this cake, you will need to have a certain amount of information, or data. So you begin to gather information through online recipes and hand-me-down cookbooks. But wait- not all the information you acquired is relevant for us, right?

So you process the information, sort of filter it, to remove the excess noise and have only what’s necessary in your head. Then you take the ingredients and apply a specific process to it which has been created solely for this cake; this process cannot be used to create a grilled cheese per say because the process is dependent on the ingredients that feed it. You put the cake in the oven for 1.5 hours, watch some Netflix, and take it out when it’s ready.

The cake is okay, not great, and you know next time you make this cake you might add a little less sugar, a little more vanilla or way more butter; I know that even a small adjustment for any of the ingredients will change the cake, our goal and outcome, tremendously.

So we keep trying to make the perfect cake, over and over, playing with the ingredients until you serve me the most delicious slice of cake these sweet lips have ever tasted.

It’s a fabulous world, darling

Welcome to the fabulous world that is machine learning.

Machine learning is about machines that are learning to perform a certain task or predict the solution to a specific problem. Learning requires information, or data. Data is all around us, we’re creating more and more of it everyday. Data must make sense, must have meaning, and must have context. That’s why we have something called features. Features, or variables, are the categories we use to label our data so we know what we’re trying to analyze.

Ok. So we have a problem/task which is making a cake.

We have data which is in this case is the measurements and amounts of all the stuff that’s supposed to go in the cake.

We have the features, which are the ingredients like sugar, milk, eggs, etc.

Next we have the process, right? This is sort of like something called an algorithm, a sort of processing equation or a set of rules that is chosen according to the data and problem at hand.

Now it gets a little tricky. When you apply your data to your algorithm, it results in something called a model. A model is built by the algorithm. The model is what you use to get a prediction.

You better learn it before you earn it

So you completed the task. Great. It took a lot of mind power and energy to make that cake and it wouldn’t be too efficient if you couldn’t leverage that whole experience and learn from it for next time, no? For example, a small child touches a hot stove and burns themselves. Next week they touch the stove again. And they repeat this pattern until their brain finally is able to learn that an object that is shaped like a stove with a red top is hot and you shouldn’t touch it.

Keyword is LEARN. When your brain learns something, like the route you take to work everyday, our actions, or outputs become sort of automated. You get to a point where you don’t need to pull and process all that data to make that decision. Your brain will automatically know what to do based on prior learnings.

Let’s summarize. Through the process I’ve described, you’re able to leverage the power of historical data to perform tasks or solve problems in an automated manner.

Fantastic. You’ve just learned what machine learning is! Not really, but you’re starting to get the picture.

Conclusion

Machine learning gets much more complex than this and we’ll definitely touch on training and test data in the next post in addition to delving more into the importance of algorithms and models. But for now, you’ve understood the basic concept of what machine learning is and the involved stakeholders. Define your problem, gather your data, sort your data and extract only the most relevant features, apply it to an algorithm, and produce your model that can help you get predictive analytics.

Until next time!