I have a random variable estimated over time by an online algorithm. I have the mean and variance of the random Gaussian variable at every step t. I expect the time series to have sudden shifts. What is the best way to estimate if the my estimated variable has shifted beyond a known threshold using t and t + 1 estimates?
Thank you.
Algorithm to detect periods with significant trends discusses detecting unusual activity in a time series . Detecting that the most recent observation is different from the expected requires an expected. If the most recent value is detected as being exceptional there are many plausible/logical reasons why this is so. It is impossible to precisely pinpoint the one "correct reason" with 1 data point.
