Table 9 Results comparison with the R outliers package
From: Contextual anomaly detection framework for big sensor data
Concept | Discussion |
|---|---|
Running time | A major difference between the work presented in this paper and the R outliers package is in the running time of both algorithms. One of the major design considerations for the proposed work was the ability to run in real-time. The R outliers package runs in batch over the entire set of data and runs on the order of minutes to completion, whereas our work runs on the order of seconds. |
Anomalies found | The work presented in this paper and R outliers package detected a number of equivalent anomalies. The huge list of anomalies pertaining to the failed sensors identified apriori by Powersmiths were found by both detectors. Further, the work presented in this paper detected all of the anomalies detected by the R outliers package, including those that were contextually cleared. |