from sklearn.pipeline import Pipeline
# project-specific helper libraries
import icu_config
import tests
######################################################################
# globals
######################################################################
RANDOM_SEED = 42 # seed for RepeatedStratifiedKFold
EPS = 10 * sys.float_info.epsilon
NRECORDS = 100 # number of patient records
FEATURES_TRAIN_FILENAME, LABELS_TRAIN_FILENAME = \
icu_config.get_filenames(nrecords=NRECORDS)
METRICS = [
"accuracy", "auroc", "f1_score", "sensitivity", "specificity", "precision"
] # sensitivity = recall
######################################################################
# functions
######################################################################
def score(y_true, y_score, metric='accuracy'):
"""
Calculates the performance metric based on the agreement between the
true labels and the predicted labels.
from sklearn.utils import resample
from sklearn.pipeline import Pipeline
# project-specific helper libraries
import icu_config
from icu_practice import score, METRICS
import classifiers
######################################################################
# globals
######################################################################
NRECORDS = 2500 # number of patient records
FEATURES_TRAIN_FILENAME, LABELS_TRAIN_FILENAME, \
FEATURES_TEST_FILENAME, LABELS_TEST_FILENAME = \
icu_config.get_filenames(nrecords=NRECORDS, test_data=True)
######################################################################
# functions
######################################################################
def get_test_scores(clf, X, y, n_bootstraps=1, metrics=['accuracy']):
"""
Estimates the performance of the classifier using the 95% CI.
Parameters
--------------------
clf : estimator object
This is assumed to implement the scikit-learn estimator interface.
The estimator must already be fitted to data.