ML Model Development Study Guide — MLA-C01

ML Model Development – MLA-C01 Study Guide

This domain accounts for 26% of the exam and covers choosing algorithms, training, tuning hyperparameters, and — crucially — evaluating models with the right metric. The metric-selection questions (precision vs recall vs F1 vs RMSE) are the highest-value and most reliably tested topic here, because they encode the business trade-off. Learn what each metric means and when it matters.

Choosing Algorithms

Match the algorithm to the problem:

• Tabular classification/regression — XGBoost (a SageMaker built-in) is a strong default: excellent results on structured data with modest tuning.
• Images — convolutional neural networks / built-in image algorithms.
• Text/NLP — sequence models or foundation models.
• Prebuilt/foundation models — SageMaker JumpStart offers models and solutions you can deploy or fine-tune quickly.

SageMaker offers built-in algorithms, script mode (bring your own training script), and BYO containers.