Precision

INFO

Measures the proportion of correctly predicted positive observations to the total predicted positives.

How It Works

$$\text{Precision}=\frac{TP}{TP+FP}$$

• TP: Correctly predicted positive cases
• FP: Incorrectly predicted positive cases

Precision focuses on how many predicted positives are actually correct.

What to Look For

• High precision = fewer false positives
• Important when false alarms are costly
• Often traded off against recall

Application Models

• Convolutional Neural Network (CNN)
• Deep Neural Network (DNN)
• Transformer