Quantization and Fudge Factors

Due to quantization, irregular measurement intervals, and round-off error, you should allow for some error when configuring the duration and level. For example, if your scheduled test interval is one minute and you want to have a better chance of having at least five items in a five minute window, set the duration to more than five minutes.

Things to consider:

• More failed cases for first few weeks: For first few weeks, the process is learning and setting the threshold by collecting data on regular basis. The more data that Hawkeye collects, more accurate is the threshold value.
• More Outliers: Outlier window moves from left to right and there are chances that outlier windows overlap for same failed cases.