Performs moving band outlier calculation.
The function computes a band in a time window, based on the percentiles within this time window and if a certain point is beyond this band (including a user defined tolerance) the value is identified as an outlier. The function returns 1 if outlier is detected
Syntax
MovingOutlier(expression, # of data points before, # of data points after, low percentile value,high percentile value, tolerance value)
- expression: signal expression
- # of data points before - integer number of points in the signal expression before the current date in the aggregation range
- # of data points after - integer number of points in the signal expression after the current date in the aggregation range
- low percentile value
- high percentile value
- tolerance value
Example
MovingOutlier("produced oil per time increment" in "STB", 3, 3, 20, 80, 1)
As you see in the following table the MovingBandOutlier with 3 data points back and 3 data points forward in time produces following output based on the 'original signal' as input:
Date/Time |
Original Signal |
MovingBandOutlier |
01.02.2007 00:00 |
14 |
1 |
01.03.2007 00:00 |
14 |
1 |
01.04.2007 00:00 |
14.86826 |
0 |
01.05.2007 00:00 |
14.66342 |
1 |
01.06.2007 00:00 |
14 |
1 |
01.07.2007 00:00 |
14.69327 |
1 |
01.08.2007 00:00 |
14.19843 |
1 |
01.09.2007 00:00 |
14.73172 |
1 |