I have an on-going periodic time series of either zeros or ones, nothing else. Mostly I get zeros, but the probability of getting ones varies over time.
What is the best way to estimate the probability of the next time period being one? Would an exponentially weighted moving average be the best method?