Skip to main content

📊 Evaluation Metrics

Evaluation Metrics are for humans to understand if the model is actually good.

🎯 The Accuracy Trap​

If 99 images are dogs and 1 is a cat, a model that always blindly guesses "Dog" gets 99% accuracy! It looks amazing, but is useless.

Instead, we use Precision, Recall, and the F1-Score!

🐍 Python Implementation​

from sklearn.metrics import classification_report

# 0 = Normal Email, 1 = Spam
true_emails = [0, 0, 0, 0, 0, 0, 0, 0, 0, 1] 
# Model just lazily guessed "0" for everything!
predictions = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0] 

# Let's see the detailed report
print(classification_report(true_emails, predictions, zero_division=0))