Quite often we see these words generously thrown out. Often it is in the right context, but sometimes not. To make it work we need data or input. The more the better, which means that predictions can be more accurate. Machine learning is not a new thing, it is all about statistics. Statistics math or, science if you will, is what drives machine learning. There are a number of basic algorithms that we use.
- Classification – Is this a photo of a car?
- Anomaly detection – Is the jet engine about to fail?
- Regression – How many phones will my company produce next month?
- Clustering – Which customers by what model of our phones?
- Reinforcement learning – What action should my drone take next to not hit the tree?
Depending on what questions about machine learning you want answers for, you apply one of the five algorithms. It is important that your question is specific. You do not want your algorithm to be able to answer with a too general answer, e.g. ‘Will it rain?’.