I have a set of data which has daily and seasonal cycle and expands for over ten years. I need to remove erroneous values using the rolling average of 5 points at a time not within +/- 3 standard deviations from the average. I might end up using more points for the rolling average (maximum of 24 values which represent the whole day) till I get something that just eliminates the wrong values. Also the +/- 3 standard deviations might be replaced by +/- 2 standard deviations as it depends on my work.
Essentially I want to remove outliers based on a rolling average principle (averaging the whole sample would dampen the cycles I have)….
Can anyone suggest how to do this in SPSS?
Ideally, the method should show me the points it identifies as outliers before just deleting them.