support for correlation of event time point processes

[sbenthall]

Much of the data we collect with bigbang is, in raw form, of the form of objects with event times in continuous time.

Technically, this is not time series data, but point process data. We get time series data from various methods of discretization of the time line and aggregation of events.

There are many ways to parameterize metrics for the correlation between point processes and time series data. Some of these are more indicative of causation than others.

Here's are handy resources with some explanation: https://stats.stackexchange.com/questions/69896/measuring-correlation-of-point-processes https://en.wikipedia.org/wiki/Granger_causality https://towardsdatascience.com/four-ways-to-quantify-synchrony-between-time-series-data-b99136c4a9c9

Most of the algorithms for these have probably been written in other packages. But we probably could figure out which are most useful for the kinds of questions we ask in BigBang, include them as a dependency, and demonstrate their use in a notebook.