Merge branch 'develop' into issue214

Author: janvanrijn

openml/runs/functions.py

@@ -5,13 +5,15 @@
 import numpy as np
 import warnings
 import sklearn
+import time
+from sklearn.model_selection._search import BaseSearchCV
 
 from ..exceptions import PyOpenMLError
 from .. import config
 from ..flows import sklearn_to_flow, get_flow, flow_exists
 from ..setups import setup_exists
 from ..exceptions import OpenMLCacheException, OpenMLServerException
-from ..util import URLError
+from ..util import URLError, version_complies
 from ..tasks.functions import _create_task_from_xml
 from .._api_calls import _perform_api_call
 from .run import OpenMLRun, _get_version_information
@@ -155,6 +157,7 @@ def _run_task_get_arffcontent(model, task, class_labels):
     X, Y = task.get_X_and_y()
     arff_datacontent = []
     arff_tracecontent = []
+    user_defined_measures = defaultdict(lambda: defaultdict(dict))
 
     rep_no = 0
     # TODO use different iterator to only provide a single iterator (less
@@ -170,20 +173,41 @@ def _run_task_get_arffcontent(model, task, class_labels):
             testY = Y[test_indices]
 
             try:
+                # for measuring runtime. Only available since Python 3.3
+                if version_complies(3, 3):
+                    modelfit_starttime = time.process_time()
                 model_fold.fit(trainX, trainY)
+
+                if version_complies(3, 3):
+                    modelfit_duration = (time.process_time() - modelfit_starttime) * 1000
+                    user_defined_measures['usercpu_time_millis_training'][rep_no][fold_no] = modelfit_duration
+
+                if isinstance(model_fold, sklearn.model_selection._search.BaseSearchCV):
+                    arff_tracecontent.extend(_extract_arfftrace(model_fold, rep_no, fold_no))
+                    model_classes = model_fold.best_estimator_.classes_
+                else:
+                    model_classes = model_fold.classes_
             except AttributeError as e:
                 # typically happens when training a regressor on classification task
                 raise PyOpenMLError(str(e))
-
+            
             # extract trace
             if isinstance(model_fold, sklearn.model_selection._search.BaseSearchCV):
                 arff_tracecontent.extend(_extract_arfftrace(model_fold, rep_no, fold_no))
                 model_classes = model_fold.best_estimator_.classes_
             else:
                 model_classes = model_fold.classes_
 
+            if version_complies(3, 3):
+                modelpredict_starttime = time.process_time()
+            
             ProbaY = model_fold.predict_proba(testX)
             PredY = model_fold.predict(testX)
+            if version_complies(3, 3):
+                modelpredict_duration = (time.process_time() - modelpredict_starttime) * 1000
+                user_defined_measures['usercpu_time_millis_testing'][rep_no][fold_no] = modelpredict_duration
+                user_defined_measures['usercpu_time_millis'][rep_no][fold_no] = modelfit_duration + modelpredict_duration
+
             if ProbaY.shape[1] != len(class_labels):
                 warnings.warn("Repeat %d Fold %d: estimator only predicted for %d/%d classes!" %(rep_no, fold_no, ProbaY.shape[1], len(class_labels)))
 
@@ -200,7 +224,6 @@ def _run_task_get_arffcontent(model, task, class_labels):
     else:
         arff_tracecontent = None
         arff_trace_attributes = None
-
     return arff_datacontent, arff_tracecontent, arff_trace_attributes
 
 
@@ -423,7 +446,7 @@ def _get_cached_run(run_id):
         run_file = os.path.join(run_cache_dir,
                                 "run_%d.xml" % int(run_id))
         with io.open(run_file, encoding='utf8') as fh:
-            run = _create_task_from_xml(xml=fh.read())
+            run = _create_run_from_xml(xml=fh.read())
         return run
 
     except (OSError, IOError):

openml/runs/run.py

@@ -149,10 +149,17 @@ def _create_description_xml(self):
 
         openml_param_settings = OpenMLRun._parse_parameters(self.model, downloaded_flow)
 
+        # as a tag, it must be of the form ([a-zA-Z0-9_\-\.])+
+        # so we format time from 'mm/dd/yy hh:mm:ss' to 'mm-dd-yy_hh.mm.ss'
+        # well_formatted_time = time.strftime("%c").replace(
+        #     ' ', '_').replace('/', '-').replace(':', '.')
+        # tags = run_environment + [well_formatted_time] + ['run_task'] + \
+        #     [self.model.__module__ + "." + self.model.__class__.__name__]
         description = _to_dict(taskid=self.task_id, flow_id=self.flow_id,
                                setup_string=_create_setup_string(self.model),
                                parameter_settings=openml_param_settings,
                                error_message=self.error_message,
+                               detailed_evaluations=self.detailed_evaluations,
                                tags=self.tags)
         description_xml = xmltodict.unparse(description, pretty=True)
         return description_xml
@@ -241,7 +248,7 @@ def _get_version_information():
     return [python_version, sklearn_version, numpy_version, scipy_version]
 
 
-def _to_dict(taskid, flow_id, setup_string, error_message, parameter_settings, tags):
+def _to_dict(taskid, flow_id, setup_string, error_message, parameter_settings, tags=None, detailed_evaluations=None):
     """ Creates a dictionary corresponding to the desired xml desired by openML
 
     Parameters
@@ -268,11 +275,17 @@ def _to_dict(taskid, flow_id, setup_string, error_message, parameter_settings, t
     if error_message is not None:
         description['oml:run']['oml:error_message'] = error_message
     description['oml:run']['oml:parameter_setting'] = parameter_settings
-    description['oml:run']['oml:tag'] = tags  # Tags describing the run
-    # description['oml:run']['oml:output_data'] = 0;
-    # all data that was output of this run, which can be evaluation scores
-    # (though those are also calculated serverside)
-    # must be of special data type
+    if tags is not None:
+        description['oml:run']['oml:tag'] = tags  # Tags describing the run
+    if detailed_evaluations is not None:
+        description['oml:run']['oml:output_data'] = dict()
+        description['oml:run']['oml:output_data']['oml:evaluation'] = list()
+        for measure in detailed_evaluations:
+            for repeat in detailed_evaluations[measure]:
+                for fold, value in detailed_evaluations[measure][repeat].items():
+                    current = OrderedDict([('@repeat', str(repeat)), ('@fold', str(fold)),
+                                           ('oml:name', measure), ('oml:value', str(value))])
+                    description['oml:run']['oml:output_data']['oml:evaluation'].append(current)
     return description
 
 

openml/util.py

@@ -12,5 +12,15 @@ def is_string(obj):
     except NameError:
         return isinstance(obj, str)
 
+def version_complies(major, minor=None):
+    version = sys.version_info
+    if version[0] > major:
+        return True
+    if version[0] < major:
+        return False
+    # version == major
+    if minor is None or version[1] >= minor:
+        return True
+    return False
 
 __all__ = ['URLError', 'is_string']