When we want to see how well supervised learning models work, we look at some important numbers. Here’s a simple breakdown of the key ones:

1. Accuracy: This is one of the easiest measures to understand. It tells us how many times the model made the right choice out of all the choices it made. While it gives a quick picture of performance, it can be tricky if the data is not balanced.

2. Precision: This number helps us understand how good the model is at making positive predictions. It’s calculated by looking at how many true positive predictions were made compared to all positive predictions. The formula is: $\text{Precision} = \frac{True Positives (TP)}{True Positives (TP) + False Positives (FP)}$

3. Recall (Sensitivity): Recall shows how well the model finds all the relevant instances. The formula for this is: $\text{Recall} = \frac{True Positives (TP)}{True Positives (TP) + False Negatives (FN)}$

4. F1 Score: The F1 Score combines both precision and recall to give us a balanced view. It’s calculated using this formula: $F1 = 2 \cdot \frac{\text{Precision} \cdot \text{Recall}}{\text{Precision} + \text{Recall}}$

5. ROC-AUC: This stands for Receiver Operating Characteristic and its area under the curve (AUC). It’s really useful for seeing how well a model performs at different settings.

Each of these numbers gives us a different picture of how well our model is doing.
This is super important when we want to evaluate any model!