We are recording very order we receive as an event. What I'd like to do is get a count every 15 minutes real time of how many orders have come in and based on the last three weeks to that day see if there was a deviation.
E..g 4:45pm Thurs Sept 4th would be compared to Aug 28th, 21st and 14th on the graph. Also we'd like to measure the std dev to see if it is +/- 1 and alert if it is.
I know we can do this in splunk but I can't seem to 1. Get the timechart/timewrap done right and 2 tell Splunk how to compare specific days of the prior 3 weeks.
Of course this is going to sound like a shameless plug, but honestly, the easiest way to do this is with the Prelert Anomaly Detective app.
Here's an example of Anomaly Detective being used against data that are "orders" - which, probably like your data, is cyclical (periodic) in nature:
By the way, don't get caught up in trying to use standard deviation as your approach to express anomalousness. Standard deviation assumes that the data samples (in this case, "counts of errors") conforms to a nice, symmetrical Gaussian Bell curve. In most cases, counts of things are better modeled by Poisson curves. Anomaly Detective automatically figures out the best statistical model for your data to maximize accuracy and minimize false alerting.
I edited your question to best reflect your use case. I just wanted to clarify that since you want the search to run every 15 minutes, you want the results, for example, from 4:45pm Thurs Sept 4th to be compared with 4:45pm Aug 28th, 4:45pm Aug 21st, 4:45pm Aug 14th correct? You said you wanted the results "on the last three weeks to that day", but you didn't mention time.