Amazon SageMaker Pipelines is a service for building and automating end-to-end ML workflows — data prep, training, evaluation, and deployment — as a CI/CD pipeline. MLA-C01 covers orchestrating ML workflows with it and Step Functions.
Amazon SageMaker is AWS's managed platform for building, training, tuning, deploying, and monitoring machine-learning models.
Feature engineering is the process of transforming raw data into input features that improve model performance, including encoding, scaling, and handling missing values.
Amazon SageMaker Feature Store is a repository for storing, sharing, and serving curated ML features for training and inference consistently.
Hyperparameter tuning is the process of searching for the hyperparameter values that produce the best model performance, which SageMaker automates with automatic model tuning.
Model training is the process of fitting an algorithm to prepared data to learn patterns, run in SageMaker as managed training jobs.
Overfitting is when a model learns the training data too closely, including noise, and generalizes poorly to new data.
Model evaluation is measuring how well a trained model performs using metrics appropriate to the task, such as accuracy, precision/recall, F1, AUC, or RMSE.
The Amazon SageMaker Model Registry is a catalog for versioning models, managing approval status, and promoting them to deployment.
A model endpoint is a deployed model that serves inference requests, which SageMaker offers as real-time, serverless, asynchronous, or batch options.
Batch transform is a SageMaker inference option that runs predictions on a whole dataset at once rather than serving live requests.
Amazon SageMaker Pipelines is a service for building and automating end-to-end ML workflows — data prep, training, evaluation, and deployment — as a CI/CD pipeline.
Amazon SageMaker Model Monitor detects data quality issues, drift, and bias in deployed models by comparing live traffic against a baseline.
Data drift is a change in the statistical distribution of production input data compared to training data, which can degrade model accuracy over time.