CompTIA DataX DY0-001 (V1) Practice Question

The correct answer is Local Outlier Factor (LOF). The LOF algorithm is specifically designed to identify local outliers by comparing the density of a data point to the densities of its neighbors. This local comparison makes it highly effective for datasets where different clusters have different densities, as it can identify a point as an outlier if it is in a sparser region than its immediate neighbors, regardless of the global data distribution.

• DBSCAN is incorrect because, while it is a density-based algorithm that can find outliers, the standard version uses global density parameters (epsilon and MinPts). This makes it struggle to correctly identify outliers in datasets with clusters of varying densities, as a single set of parameters is often not suitable for both sparse and dense regions.
• Mahalanobis Distance is incorrect because it measures the distance of a point from the center of a distribution, accounting for covariance. However, it assumes the data follows a multivariate normal (elliptical) distribution and is less effective for datasets with complex, non-elliptical cluster structures.
• Z-Score is incorrect because it is a univariate method, meaning it assesses one feature at a time. It does not account for the correlations between variables in a multivariate dataset and would fail to identify outliers that are only apparent when considering multiple dimensions together.