filtering the dataset (was: Compute statistics for part of a sequence)

[mlmeijerink]

For some applications, it may be useful that statistics can be computed for only part of a sequence and not for the whole sequence, while basing them on the whole sequence. This is currently available in the remstatsMW function, but should be made more intuitive (e.g., feed the whole sequence and set a start and a stop event).

[SNAnalyst]

I think we should probably define a separate function that takes the appropriate slice from the rem data. The slice can be taken based on start time/end time or start event/end event. Additionally, we could filter events based on specific senders/receivers/senders-or-receivers or subgroups/attributes of actors.

The resulting data can then be used in the remstats function.

Actually, we might want to move this function into a separate package, perhaps as part of the remdata package, which could then contain datasets + data manipulation functions. By making this into a separate function, it can also be used to generate input to the visualization functions, summary functions, etc.

[mlmeijerink]

In remstats version 1.0.0, it is possible to filter the relational event history for which statistics need to be computed based on a start event/end event. This option is part of the remstats function, so that statistics can be computed for a slice of the data while still including past events that are not in this slice but are in the edgelist that is submitted.

When remdata is introduced, it may be decided to remove this functionality from remstats and let remstats call the appropriate function in remdata to handle this.

Filtering on specific senders/receivers, etc. is not yet implemented.