refactored ModelResults into separate classes for different BackEnds

bambi/backends.py

@@ -2,7 +2,7 @@
 from bambi.external.six import string_types
 import numpy as np
 import warnings
-from bambi.results import ModelResults
+from bambi.results import PyMC3ModelResults
 from bambi.priors import Prior
 import theano
 try:
@@ -122,11 +122,11 @@ def run(self, start=None, find_map=False, **kwargs):
             find_map (bool): whether or not to use the maximum a posteriori
                 estimate as a starting point; passed directly to PyMC3.
             kwargs (dict): Optional keyword arguments passed onto the sampler.
-
+        Returns: A PyMC3ModelResults instance.
         '''
         samples = kwargs.pop('samples', 1000)
         with self.model:
             if start is None and find_map:
                 start = pm.find_MAP()
             self.trace = pm.sample(samples, start=start, **kwargs)
-        return ModelResults(self.spec, self.trace)
+        return PyMC3ModelResults(self.spec, self.trace)

bambi/results.py

@@ -1,36 +1,38 @@
 import pandas as pd
 import pymc3 as pm
+from abc import abstractmethod, ABCMeta
 
 
 class ModelResults(object):
 
-    def __init__(self, model, trace):
+    __metaclass__ = ABCMeta
+
+    def __init__(self, model):
 
         self.model = model
         self.terms = list(model.terms.values())
-        self.trace = trace
         self.diagnostics = model._diagnostics
         self.n_terms = len(model.terms)
+
+    @abstractmethod
+    def plot(self):
+        pass
+
+    @abstractmethod
+    def summary(self):
+        pass
+
+
+class PyMC3ModelResults(ModelResults):
+
+    def __init__(self, model, trace):
+
+        self.trace = trace
         self.n_samples = len(trace)
-        self._fixed_terms = [t.name for t in self.terms if t.type_=='fixed']
-        self._random_terms = [t.name for t in self.terms if t.type_=='random']
-
-    def _select_samples(self, fixed, random, names, burn_in):
-        trace = self.trace[burn_in:]
-        if names is not None:
-            names = []
-            if fixed:
-                names.extend(self._fixed_terms)
-            if random:
-                names.extend(self._random_terms)
-        return trace, names
-
-    def plot_trace(self, burn_in=0, fixed=True, random=True, names=None,
-                   **kwargs):
-        trace, names = self._select_samples(fixed, random, names, burn_in)
-        return pm.traceplot(trace, varnames=names, **kwargs)
-
-    def summary(self, burn_in=0, fixed=True, random=True, names=None, **kwargs):
-        trace, names = self._select_samples(fixed, random, names, burn_in)
-        return pm.summary(trace, varnames=names, **kwargs)
 
+    def plot(self, burn_in=0, names=None, **kwargs):
+        return pm.traceplot(trace[burn_in:], varnames=names, **kwargs)
+
+    def summary(self, burn_in=0, fixed=True, random=True, names=None,
+                **kwargs):
+        return pm.summary(trace[burn_in:], varnames=names, **kwargs)