paoloalborno
Joined: 11 Feb 2015, 17:26 Posts: 3
|
![Reply with quote Reply with quote](./styles/avalon/imageset/en/icon_post_quote.gif)
Hi everybody, we are developing a an interesting algorithm for multi event class synchronization. Algorithm features:
Problem: We want to measure synchronization and leadership between two time series, that contains event that belong to different classes (A,B… etc).
Event mapping: Classes of events like A,B.. are now codified by numbers in the TS: A = 1, B = 2 …
a) Multi event class: calculates synchronization and leadership between the two time series, in which different classes of events has been defined. Every event class is identified uniquely by and integer in the TS. Example: every event class is identified uniquely by an integer in the TS. A time series with 3 classes is : {....0 1 0 2 3 0 1 0 3….}.
b) Nominal sets: this modality requires the definition of sets of event classes. Every set is composed by at least one class, and the synchronization is calculated between all the events that are part of the set itself. The output is a vector containing the computed sync index for each set that has been defined. An example two sets is {1,2,3} and {4,5,6} .
c) Sequence of events: this modality uses sets of events classes, but events should occur following a precise order (a sequence of events) otherwise they will be not considered as valid. To consider the events as part of the sequence, an inter_event time should be defined. The output will be a vector containing the computed sync index for each sequence that has been defined. Nominal sets as {1,4,5} and {2,3} will be considered as sequences.
|