Confusion Matrix

Published 2026-02-10 in Deep Learning

A Confusion Matrix is one of the most important evaluation tools used in classification problems . It provides a detailed breakdown of how a machine learning model’s predictions compare with the actual ground truth labels. Unlike accuracy alone, a confusion matrix helps us understand where the model is making mistakes and what type of errors occur . The image above represents a binary classification problem, such as disease detection , where outcomes are classified as positive or negative. Structure of the Confusion Matrix The confusion matrix is a 2×2 table that compares: Predicted Labels (rows) Actual / Ground Truth Labels (columns) Ground Truth Labels Condition Positive (CP): The actual condition is present (e.g., patient has disease) Condition Negative (CN): The actual condition is absent (e.g., patient has no disease) Predicted Labels Test Outcome Positive (TOP): Model predicts positive Test Outcome Negative (TON): Model predicts negative The Four Outcomes Explained 1. True Positive (TP) The model predicts positive , and the actual condition is also positive . Example: The model correctly identifies a patient who has the disease. 2. False Positive (FP) The model predicts…

More Deep Learning articles · All collections · Practice challenges