search

Eden-Kramer-Lab / replay_trajectory_classification

State space models for decoding hippocampal trajectories and determining their type using sorted or clusterless data
MIT License
43 stars 13 forks source link

spiking likelihood #34

Closed samdeoxys1 closed 3 months ago

samdeoxys1 commented 3 months ago

Is it true that the spikes are expected to be 0/1, instead of any integer, when using the spiking_likelihood_kde? Does that mean if my bin is large and there are more than one spikes per bin, the result is not accurate?

edeno commented 3 months ago

For spiking_likelihood_kde the results should be fine with larger time bins. A count vector would be fine. If you were using the clusterless algorithms in this package as implemented, then there would be more of a problem with larger time bins (something I fixed in a forthcoming package but not here).