Question 27
UnclassifiedWhat are t-SNE and UMAP primarily used for in data analysis and machine learning?
Correct answer: B
Explanation
t-SNE and UMAP are used to map high-dimensional data into a lower-dimensional space while preserving local structure, which is why they are described as "non-linear dimensionality reduction" methods. They are also widely used for visualization because the reduced embeddings make clusters and patterns easier to see in 2D or 3D.
Why each option is right or wrong
A. Predictive modeling and supervised classification of new samples
B. Non-linear dimensionality reduction and visualization
t-SNE (van der Maaten & Hinton, 2008) and UMAP (McInnes, Healy & Melville, 2018) are both algorithms designed to embed high-dimensional observations into 2D or 3D while preserving neighborhood relationships, which is the defining feature of non-linear dimensionality reduction. In practice, they are used to make latent clusters, separations, and manifold structure visible in plots; they are not primarily classification or regression methods, and there is no fixed threshold or decision rule involved.
C. Hyperparameter tuning for training machine learning models
D. Anomaly detection only, without reducing dimensionality