Taking observations or attributes out of a data set prior to data modeling is called _________, which is part of the Data Preparation step of CRISP-DM.

• A. Elimination
• B. Pruning
• C. Reduction
• D. Selection

Answer: (C)

Reducing the dataset before modeling means trimming away parts that aren’t needed so the data is smaller and easier to work with, while still keeping the information that matters for the model. In CRISP-DM, the Data Preparation phase includes cleaning and shaping data, and this often involves reducing data size or dimensionality—removing some observations (rows) or attributes (columns) to simplify the dataset and improve modeling performance. That broad action is exactly what “reduction” captures, making it the best fit for this description.

Other terms like elimination or pruning convey similar ideas of removing things, but they aren’t the standard umbrella term used to describe this pre-modeling simplification in CRISP-DM. Selection might refer to choosing a subset of attributes rather than removing parts of the data, which is more about picking what to keep rather than reducing the overall dataset. Reducing data size or dimensionality is the most general and appropriate label for taking out observations or attributes prior to modeling.