3 Measuring Performance
As we build systems to predict, classify, and otherwise find patterns in our data, we’ll need some way to discuss how well they’re doing their job. We use a variety of numerical measurements for just this purpose, which we collectively call performance metrics. They’ve been designed to enable us to carefully describe what the system is doing right, and more importantly, when the system gets the wrong answers, specifically how those answers are wrong. These tools are the keys to interpreting any system’s results.
Our metrics are based on probability, or how likely it is that we’ll see different types of results. So ...
Get Deep Learning now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
