To do list

[StochasticBiology]

Remember that pulling HyperTraPS parameterisations from HyperHMM fit doesn't guard at all against overfitting. Perhaps we can regularise the derived parameterisation natively?

• expand model conversion for Mk, LAU, DAGs, etc -- in progress
• ensure likelihoods are consistently reported across types
• ensure predictions can consisently be made across types
• Handle timings for continuous-time data
• Mk model can handle uncertainty, put priors over states
• Not sure uncertainty is handled correctly in simple demo example
• Work on integrating other plot types
• Build in clustering ideas from CHHMM?