An ML engineer uses Amazon QuickSight anomaly detection to flag very high or very low machine operating temperatures compared to normal. Severity is set to "Low and above" and Direction is set to "All." If the engineer changes Direction to "Lower than expected," what effect will be observed in anomaly detection results?

Changing the anomaly detection direction from "All" to "Lower than expected" will reduce the number of conditions under which an anomaly is flagged. Previously, QuickSight would identify anomalies for both unusually high and unusually low temperatures. Now, it will only flag temperatures that are significantly lower than the expected range. This narrower focus means fewer events will meet the criteria for an anomaly, leading to less frequent anomaly identification. Consequently, recall will decrease because the system will miss anomalies that are higher than expected, which it previously would have caught.