def buildFirstClassificationModel(self):
global features, feature_names, target, target_names
labels = target_names[target]
is_setosa = labels == "setosa"
# get then max petal length of setosa and min petal length of non-setosa
plength_setosa = features[is_setosa, 2]
plength_non_setosa = features[~is_setosa, 2]
maxSetosa = plength_setosa.max()
min_nonSetosa = plength_non_setosa.min()
logging.info("max plength for setosa : %f", maxSetosa)
logging.info("min plength for non-setosa : %f", min_nonSetosa)
# select Features & Labels for non Setosa:
features_nonSetosa = features[~(labels == "setosa")]
labels_nonSetosa = labels[labels != 'setosa']
is_virginica = (labels_nonSetosa == "virginica")
accuracy_results, best_fi, best_t, best_reverse = getModel(features_nonSetosa, is_virginica, feature_names)
self.drawAccuracyComparisionChart(accuracy_results)
# test the threshold on a sample
example = np.array([1.6, 2.5, 4.3, 2.6])
self.is_virginica_test(best_fi, best_t, best_reverse, example)
def cross_validation(self):
global features, feature_names, target, target_names
is_setosa = target_names[target] == "setosa"
features_wo_setosa = features[~is_setosa]
labels_wo_setosa = target_names[target][~is_setosa]
is_virginica = labels_wo_setosa == "virginica"
# set a vector of all but one True of the size of features_wo_setosa
correct = 0.0
for ei in range(len(features_wo_setosa)):
train = np.ones(len(features_wo_setosa), bool)
train[ei] = False
testing = ~train
model = getModel(features_wo_setosa[train], is_virginica[train], feature_names)
correct += np.sum(predict(model, features_wo_setosa[testing]) == is_virginica[testing])
acc = correct / float(len(features_wo_setosa))
logging.info("Accuracy on cross-validation : %f", acc)
def train_test_50(self):
global features, feature_names, target, target_names
labels = target_names[target]
is_setosa = labels == "setosa"
features_wo_setosa = features[~is_setosa]
labels_wo_setosa = labels[~is_setosa]
is_virginica = (labels_wo_setosa == "virginica")
# split the features in train and test
training = np.tile([False, True], 50)
testing = ~training
# display the resulting accuracy for both series
model_training = getModel(features_wo_setosa[training], is_virginica[training], feature_names)
accuracy_training = accuracy(model_training, features_wo_setosa[training], is_virginica[training])
accuracy_testing = accuracy(model_training, features_wo_setosa[testing], is_virginica[testing])
logging.info("Training Accuracy : %f ----Testing Accuracy = %f ", accuracy_training, accuracy_testing)
from main.ch02.utils import getModel, accuracy
data = load_iris()
features = data['data']
labels = data['target_names'][data['target']]
# We are going to remove the setosa examples as they are too easy:
is_setosa = (labels == 'setosa')
features = features[~is_setosa]
labels = labels[~is_setosa]
# Now we classify virginica vs non-virginica
is_virginica = (labels == 'virginica')
# Split the data in two: testing and training
testing = np.tile([True, False], 50) # testing = [True,False,True,False,True,False...]
# Training is the negation of testing: i.e., datapoints not used for testing,
# will be used for training
training = ~testing
model = getModel(features[training], is_virginica[training])
train_accuracy = accuracy(features[training], is_virginica[training], model)
test_accuracy = accuracy(features[testing], is_virginica[testing], model)
print('''\
Training accuracy was {0:.1%}.
Testing accuracy was {1:.1%} (N = {2}).
'''.format(train_accuracy, test_accuracy, testing.sum()))
