Time Series Outlier Detection Based on Sliding Window Prediction

Table 1

Assessment of both methods.

Truth	Detection
	Outlier	Not an outlier
Outlier	True positives or TP (A) data points that are outliers and identify as outliers	False negatives or FN (C) data points that are outliers but identify as normal
Not an outlier	False positives or FP (B) data points that are normal but identify as outliers	True negatives or TN (D) data points that are normal and identify as normal