AIExplainer
Machine Learning Intermediate 2 min read

What is a metric?

A quantitative measure used to evaluate the performance of a model or system

A metric is a way to measure how well a model or system is doing its job. It's a number that shows how good or bad the performance is, and it's used to compare different models or systems.

A metric is like a report card grade for a model or system. Just as a grade shows how well a student is doing in school, a metric shows how well a model or system is doing its job.

For example, a company might use a metric called 'accuracy' to evaluate the performance of a model that predicts customer churn. If the model is 90% accurate, that means it correctly predicts whether a customer will churn 90% of the time.

Metrics are used to evaluate the performance of models or systems, and to compare different models or systems. They are often used to optimize the performance of a model or system, by adjusting parameters or trying different approaches.

One common misconception is that a single metric is enough to evaluate the performance of a model or system. In reality, multiple metrics are often needed to get a complete picture of performance.

The use of metrics to evaluate performance dates back to the early days of statistics and quality control. In recent years, the use of metrics has become increasingly important in the field of artificial intelligence and machine learning.

measure indicator benchmark

Three products for different needs — explore what’s relevant to you.