It doesn't seem possible currently to use noise_models other than NB, yet these should be implemented in batchGLM ? Could this feature be added ?
|
if noise_model == "nb" or noise_model == "negative_binomial": |
|
from batchglm.api.models.numpy.glm_nb import Estimator, InputDataGLM |
|
else: |
|
raise ValueError('noise_model="%s" not recognized.' % noise_model) |