I need a numeric measure that determines a level of sustained activity.
Let me describe a situation in which this could be useful. Imagine that you have a Twitter feed in which you are tracking words that do not appear in the dictionary. The purpose being that you're trying to figure out if there any new hip words that people are using. So not only does a word need to be frequently used it needs to have sustained usage.
A simple way of doing this is to record summary statistics for each word. So the count of the word indicates popularity and the standard deviation of the time stamps of a word usage could indicate sustained usage of that word. Or perhaps kurtosis could?
One issue however with standard deviation is that if a word is very very popular it tends to narrow the standard deviation, so that might not be the best indicator of usage.
If a word is used very infrequently the standard deviation is bloated and would overemphasize usage.
Any recommendations for a metric or measure?