Modeling Domain

SageMaker ships built-in algorithms, and this domain (36% of the exam) tests matching problem → algorithm:

• Tabular classification/regression — XGBoost (the default), Linear Learner
• Sparse data, clicks, recommendations — Factorization Machines
• Clustering / dimensionality — k-Means, PCA
• Anomaly detection — Random Cut Forest; IP Insights for suspicious IPs
• Embeddings for pairs — Object2Vec (recommendation, doc similarity)
• Text — BlazingText (classification + word embeddings), Seq2Seq, LDA/NTM for topics
• Time series — DeepAR
• Vision — Image Classification (whole image), Object Detection (boxes), Semantic Segmentation (pixels)