Expose n_jobs param for relevant models, set default to n_cores

Author: acrellin

cesium_app/ext/sklearn_models.py

@@ -1,3 +1,4 @@
+import os
 import collections
 from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor
 from sklearn.linear_model import (LinearRegression, SGDClassifier,
@@ -12,6 +13,8 @@
                     'BayesianARDRegressor': ARDRegression,
                     'BayesianRidgeRegressor': BayesianRidge}
 
+N_JOBS_DEFAULT = os.cpu_count()
+
 
 def make_list(x):
     """Wrap `x` in a list if it isn't already a list or tuple.
@@ -49,7 +52,8 @@ def make_list(x):
          {"name": "bootstrap", "type": bool, "default": True},
          {"name": "oob_score", "type": bool, "default": False},
          {"name": "random_state", "type": int, "default": None},
-         {"name": "class_weight", "type": dict, "default": None}],
+         {"name": "class_weight", "type": dict, "default": None},
+         {"name": "n_jobs", "type": int, "default": N_JOBS_DEFAULT}],
      "type": "classifier",
      "url": "http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html"},
 
@@ -67,7 +71,8 @@ def make_list(x):
          {"name": "bootstrap", "type": bool, "default": True},
          {"name": "oob_score", "type": bool, "default": False},
          {"name": "random_state", "type": int, "default": None},
-         {"name": "class_weight", "type": dict, "default": None}],
+         {"name": "class_weight", "type": dict, "default": None},
+         {"name": "n_jobs", "type": int, "default": N_JOBS_DEFAULT}],
      "type": "classifier",
      "url": "http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html"},
 
@@ -84,7 +89,8 @@ def make_list(x):
          {"name": "max_leaf_nodes", "type": int, "default": None},
          {"name": "bootstrap", "type": bool, "default": True},
          {"name": "oob_score", "type": bool, "default": False},
-         {"name": "random_state", "type": int, "default": None}],
+         {"name": "random_state", "type": int, "default": None},
+         {"name": "n_jobs", "type": int, "default": N_JOBS_DEFAULT}],
      "type": "regressor",
      "url": "http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html"},
 
@@ -101,7 +107,8 @@ def make_list(x):
          {"name": "max_leaf_nodes", "type": int, "default": None},
          {"name": "bootstrap", "type": bool, "default": True},
          {"name": "oob_score", "type": bool, "default": False},
-         {"name": "random_state", "type": int, "default": None}],
+         {"name": "random_state", "type": int, "default": None},
+         {"name": "n_jobs", "type": int, "default": N_JOBS_DEFAULT}],
      "type": "regressor",
      "url": "http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestRegressor.html"},
 
@@ -122,14 +129,16 @@ def make_list(x):
          {"name": "eta0", "type": float, "default": 0.0},
          {"name": "power_t", "type": float, "default": 0.5},
          {"name": "class_weight", "type": [dict, str], "default": None},
-         {"name": "average", "type": [bool, int], "default": False}],
+         {"name": "average", "type": [bool, int], "default": False},
+         {"name": "n_jobs", "type": int, "default": N_JOBS_DEFAULT}],
      "type": "classifier",
      "url": "http://scikit-learn.org/stable/modules/generated/sklearn.linear_model.SGDClassifier.html"},
 
     {"name": "LinearRegressor",
      "params": [
          {"name": "fit_intercept", "type": bool, "default": True},
-         {"name": "normalize", "type": bool, "default": False}],
+         {"name": "normalize", "type": bool, "default": False},
+         {"name": "n_jobs", "type": int, "default": N_JOBS_DEFAULT}],
      "type": "regressor",
      "url": "http://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html"},
 

cesium_app/handlers/model.py

@@ -62,7 +62,7 @@ def _build_model_compute_statistics(fset_path, model_type, model_params,
         raise ValueError("Cannot build model for unlabeled feature set.")
     model = MODELS_TYPE_DICT[model_type](**model_params)
     if params_to_optimize:
-        model = GridSearchCV(model, params_to_optimize, n_jobs=-1)
+        model = GridSearchCV(model, params_to_optimize)
     model.fit(fset, data['labels'])
     score = model.score(fset, data['labels'])
     best_params = model.best_params_ if params_to_optimize else {}