Our algorithm considers two situations when doing this anomaly calculation. One is the change of rank score between these two selected days. If this movement between these selected days is high, it causes anomaly score to increase. The other is the position of the ranking. Anomaly score is calculated higher for those whose rank score is higher. For example, if the ranking scores of the two selected days are low, the movement will not increase any more, even if the movement is high.
Anomaly metric gives us the severity of changes in ranking between two selected dates of a keyword.
Written by Zeynep DerinozUpdated over a week ago