I’m trying to wrap my head around this whole data science thing. I’ve been playing around with some basic analytics tools, but I keep seeing these terms “machine learning” and “predictive analytics” pop up everywhere. I mean, I get the basic idea that they’re both about predicting stuff, right? But I’m confused about how they’re different.
I once tried to use a tool that claimed to use machine learning to predict customer churn, but it was honestly just as good as a simple spreadsheet. So, is machine learning really that much better than regular analytics?
Predictive analytics encompasses a range of statistical techniques, while machine learning is a subset focused on automatically learning from data. The ideal approach depends on the complexity of the problem, the data available, and the level of accuracy required.
Predictive analytics is the study of guessing what an unknown variable will be worth based on the values of other factors that are known. For instance, guessing how traffic will be in the future based on how it is now and how it has been in the past. Prescriptive analytics is the process of choosing the best values for one or more decision factors in order to improve one or more outcomes while still following certain rules. For instance, figuring out the fastest way to get to a certain place while taking into account expected traffic. Machine Learning is the process of teaching a set of variables how to work together or what their role is. In predicted analytics, machine learning is used a lot of the time. It is also used sometimes in predictive analytics.
As ML is essentially a broad category for novices, deep learning can also be categorized under it. While this isn’t always the case, Google’s ML course covers neural nets.
ML functions more like a generic code, completing tasks without the need for explicit coding.
Predictive is similar to the results of a machine learning model, such as the output of a nation for the next five years.
Prescriptive decision-making involves making predictions—or, more crucially, data-driven decisions.