--export-dref: calculate model complexity

2 files changed, 32 insertions, 0 deletions

diff --git a/lib/functions.py b/lib/functions.py
index 9e5639a..c229594 100644
--- a/lib/functions.py
+++ b/lib/functions.py
@@ -197,6 +197,9 @@ class ModelFunction:
     def eval_arr(self, params):
         raise NotImplementedError
 
+    def get_complexity_score(self):
+        raise NotImplementedError
+
     def eval_mae(self, param_list):
         """Return model Mean Absolute Error (MAE) for `param_list`."""
         if self.is_predictable(param_list):
@@ -282,6 +285,9 @@ class StaticFunction(ModelFunction):
     def eval_arr(self, params):
         return [self.value for p in params]
 
+    def get_complexity_score(self):
+        return 1
+
     def to_json(self, **kwargs):
         ret = super().to_json(**kwargs)
         ret.update({"type": "static", "value": self.value})
@@ -377,6 +383,14 @@ class SplitFunction(ModelFunction):
                 ret += 1
         return ret
 
+    def get_complexity_score(self):
+        if not self.child:
+            return 1
+        ret = 0
+        for v in self.child.values():
+            ret += v.get_complexity_score()
+        return ret
+
     def to_dot(self, pydot, graph, feature_names, parent=None):
         try:
             label = feature_names[self.param_index]
@@ -537,6 +551,9 @@ class CARTFunction(SKLearnRegressionFunction):
     def get_max_depth(self):
         return self.regressor.get_depth()
 
+    def get_complexity_score(self):
+        return self.get_number_of_leaves()
+
     def to_json(self, feature_names=None, **kwargs):
         import sklearn.tree
 
@@ -588,6 +605,9 @@ class LMTFunction(SKLearnRegressionFunction):
     def get_number_of_leaves(self):
         return len(self.regressor._leaves.keys())
 
+    # def get_complexity_score(self):
+    #    FIXME
+
     def get_max_depth(self):
         return max(map(len, self.regressor._leaves.keys())) + 1
 
@@ -635,6 +655,9 @@ class XGBoostFunction(SKLearnRegressionFunction):
             ret.append(self._get_max_depth(child))
         return 1 + max(ret)
 
+    def get_complexity_score(self):
+        return self.get_number_of_leaves()
+
 
 # first-order linear function (no feature interaction)
 class FOLFunction(ModelFunction):
@@ -766,6 +789,9 @@ class FOLFunction(ModelFunction):
             )
             raise
 
+    def get_complexity_score(self):
+        return len(self.model_args)
+
     def to_json(self, **kwargs):
         ret = super().to_json(**kwargs)
         ret.update(
@@ -978,6 +1004,9 @@ class AnalyticFunction(ModelFunction):
             )
             return self.value
 
+    def get_complexity_score(self):
+        return len(self.model_args)
+
     def webconf_function_map(self):
         js_buf = self.model_function
         for i in range(len(self.model_args)):
diff --git a/lib/parameters.py b/lib/parameters.py
index 51c47b8..00260fc 100644
--- a/lib/parameters.py
+++ b/lib/parameters.py
@@ -642,6 +642,9 @@ class ModelAttribute:
     def to_dref(self, unit=None):
         ret = {"mean": (self.mean, unit), "median": (self.median, unit)}
 
+        if issubclass(type(self.model_function), df.ModelFunction):
+            ret["complexity"] = self.model_function.get_complexity_score()
+
         if type(self.model_function) in (
             df.SplitFunction,
             df.CARTFunction,