esem.emulator.Emulator¶
- class esem.emulator.Emulator(model, training_params, training_data, name='', gpu=0)¶
A class wrapping a statistical emulator
- training_data¶
A wrapped representation of the training data
- model¶
The underlying model which performs the emulation
- Type
ModelAdaptor
- name¶
A human-readable name for the model
- Type
str
- __init__(model, training_params, training_data, name='', gpu=0)¶
- Parameters
model (
ModelAdaptor
) – The (compiled but not trained) model to be wrappedtraining_params (
pd.DataFrame
or array-like) – The training parameters (X)training_data (
esem.wrappers.DataWrapper
orxarray.DataArray
oriris.Cube
or array-like) – The training data - the leading dimension should represent training samples (Y)name (
str
) – Human readable name for the modelgpu (
int
) – The machine GPU to assign this model to
Methods
__init__
(model, training_params, training_data)- Parameters
model (
ModelAdaptor
) – The (compiled but not trained) model to be wrapped
batch_stats
(sample_points[, batch_size])Return mean and standard deviation in model predictions over samples, without storing the intermediate predicions in memory to enable evaluating large models over more samples than could fit in memory
predict
(x, *args, **kwargs)Make a prediction using a trained emulator
train
([verbose])Train on the training data