ensemble-pkg 0.0.1

Build deployable model ensembles without refactoring your code.

ensemble

A model ensemble utility optimized for low barrier integration

TL;DR this lets you use one thing to call many things.

Examples

Define your model functions and create your ensemble:

>>> from ensemble import Ensemble
>>> def square(x):
...     return x**2
...
>>> def cube(y):
...     return y**3
...
>>> my_ensemble = Ensemble(
...     name='e1',
...     children=[function1, function2],
... )

Multiplex between functions:

>>> my_ensemble(child='square', x=2)
4
>>> my_ensemble(child='cube', y=2)
8

Call all the models in the ensemble:

>>> my_ensemble.all(x=2, y=2)
{'square': 4, 'cube': 8}

You may instead decorate your model functions with @model in order to attach them to an ensemble:

>>> from ensemble import child
>>> @child('e2')
... def func1(x):
...     return x**2
...
>>> @child('e2')
... def func2(x):
...     return x**3
...
>>> e2 = Ensemble('e2')
>>> e2.all(x=3)
{'func1': 9, 'func2': 27}

You may even attach a model to multiple ensembles! (this is one main reason ensemble is useful)

>>> @child('e2', 'e3')
... def func3(x, y):
...     return x**3 + y
...
>>> e2.all(x=2, y=3)
{'func1': 4, 'func2': 8, 'func3': 11}
>>>
>>> e3 = Ensemble('e3')
>>> e3.all(x=2, y=3)
{'func3': 11}

If you forget what models are in your ensemble, just check:

>>> e2
Ensemble(
  name='e2',
  children={
    'func1': <function func1 at 0x1024fa9d8>
    'func2': <function func2 at 0x1024faa60>
    'func3': <function func3 at 0x1024fa950>
  },
  weights=None,
)
>>> e3
Ensemble(
  name='e3',
  children={
    'func3': <function func3 at 0x1024fa950>
  },
  weights=None,
)

In the above example, ensemble e2 contains func1, func2, and func3, while ensemble e3 contains just func3.