4.7 Model Training and Evaluation
Training = the model shrinking its mistakes step by step. Evaluation = checking scores like accuracy on fresh data.
Chapter 4: Introduction to Machine Learning (ML)
During training, the model measures how wrong it is using a loss function - basically an error score. Then an optimiser nudges its internal numbers to reduce that loss. Each full pass through the training data is called an epoch.
If you plot loss over time, it should go down. If the training loss keeps dropping but the validation loss starts going up, that's overfitting kicking in - time to stop.
For evaluation, you use metrics like accuracy (how often it's right). We'll go deeper on metrics like precision and recall in Chapter 8.
Key takeaways
Loss = how wrong the model is; training lowers it
Epoch = one pass through the training data
Watch validation loss to catch overfitting