shah-in-boots/orbs

Cyclical and rhythm based statistics

orbs's Issues

composite event data

Merge multiple events into a composite event when needed. Would need to take earliest date in that case (if just using traditional hazard modeling).

Cyclical data definition

Cyclical data will be both continuous and categorical. A type stable definition should be available as specific classes:

• continuous versus discrete/categorical variables
• stable time intervals or sampling frequencies vs. irregular sampling pattern

Harmonic regressions

This will be a complex port over from asshah4/card to pull the cosinor and harmonic regression options here, where it more likely belongs.

• Base cosinor function
• Tidyverse/tidymodel implementation of the cosinor regression
• cosinor for population level
• multiple phase evaluation (e.g. secondary tau or period elements)

competing risk datasets

Competing risk in {crr} and {crrSC} packages require specific way to format data. Should have wrapper function to generate competing risks more easily. Data should look like:

Observation	Failure	Event Type
A	0	Censor
B	0	Sudden Cardiac Death
C	1	Shark Attack

Would need to revise traditional censoring time points based on what comes earlier.

Revisioning recurrent/cyclic events

Rebranding the project to handle recurrent and repeating events in a cyclic manner. This is different than truly recurrent events with non-predictable cycle lengths (e.g. survival or recurrent event analyses).