search

tapios / risk-networks

Code for risk networks: a blend of compartmental models, graphs, data assimilation and semi-supervised learning
Other
2 stars 2 forks source link

binomial health service degree distribution #108

Closed lubo93 closed 4 years ago

lubo93 commented 4 years ago

This PR changes assign_health_workers to get the correct binomial degree distribution (i.e., the distribution that we describe in our manuscript) between health workers and hospitalized patients: size = np.random.binomial(len(viable_health_workers), self.health_workers_per_patient/len(viable_health_workers)).

I would use the binomial case as standard and not the regular degree distribution that we used previously.

odunbar commented 4 years ago

I am fine with this! Although we should update the document too, as it states patients can interact with each other and still mentions 'hospital beds'.

We do not allow patient-patient interactions in our current model (as you only become a patient if you have the virus). Otherwise we would require the health service to also assign other random nodes from categories (S,E,I,R) to the hospital (i.e to add these edges) to act as other patients which is of course possible, but we haven't felt this necessary as of yet.