問題文
A navigator is showing a heat map of hosts colored by CPU utilization. The engineer wants to find the machines that are behaving differently from their peers rather than simply the machines with the highest values. What does the navigator offer for this, and what should the engineer be careful about?
選択肢
- Changing the metric in the color-by menu until a red square appears, because red always indicates a problem and white always indicates a healthy instance, whatever metric is selected in the population view.
- Sorting the table by the colored column in descending order, because the instances that differ from their peers are by definition the ones with the most extreme values and always appear at the top of that sort whatever metric is chosen.
- Grouping by a dimension, which by itself identifies the outliers; each group is drawn as its own block and the block that looks different from the rest is the answer for that metric.
- Outlier detection, which highlights instances whose values differ significantly from the population, so the engineer should choose between deviation from the mean and deviation from the median, because the mean strategy needs a large population to be meaningful.