You're working with petabytes of data and need to make this dataset more manageable. To do this, you want to reduce the number of variables under consideration. What is the name for this process?

Dimensionality Reduction is the formal process of decreasing the number of input variables (features) while preserving as much information as possible, making petabyte-scale datasets easier to store, visualize, and process. Techniques such as Principal Component Analysis (PCA) or t-SNE project the high-dimensional data into lower-dimensional sub-spaces, thereby reducing computational complexity and mitigating the curse of dimensionality.

B. Gradient Descent is an optimization algorithm, not a feature-count reduction technique.
C. Multivariate regression is a predictive modeling method; it may use many variables but does not reduce them.
D. Data selection filters records, not the dimensionality of variables.

1. MIT 6.036 “Introduction to Machine Learning” Lecture 12 Notes, “Dimensionality Reduction”, pp. 1-3.

2. Van der Maaten & Hinton, “Visualizing Data using t-SNE”, JMLR 9, 2008, §2. https://jmlr.org/papers/volume9/vandermaaten08a/vandermaaten08a.pdf

3. Jolliffe & Cadima, “PCA: a review and recent developments”, Phil. Trans. R. Soc. A 374:20150202, 2016, p.2. https://doi.org/10.1098/rsta.2015.0202