def figures_mesa_three_class():
classifiers = utils.get_classifiers()
# Uncomment to just use MLP:
# classifiers = [AttributedClassifier(name='Neural Net', classifier=MLPClassifier(activation='relu', hidden_layer_sizes=(15, 15, 15),
# max_iter=1000, alpha=0.01, solver='lbfgs'))]
feature_sets = utils.get_base_feature_sets()
three_class_performance_summaries = []
for attributed_classifier in classifiers:
if Constants.VERBOSE:
print('Running ' + attributed_classifier.name + '...')
classifier_summary = ThreeClassClassifierSummaryBuilder.build_mesa_leave_one_out(
attributed_classifier, feature_sets)
PerformancePlotBuilder.make_bland_altman(classifier_summary, '_mesa')
PerformancePlotBuilder.make_single_threshold_histograms(
classifier_summary, '_mesa')
for attributed_classifier in classifiers:
if Constants.VERBOSE:
print('Running ' + attributed_classifier.name + '...')
classifier_summary = ThreeClassClassifierSummaryBuilder.build_mesa_all_combined(
attributed_classifier, feature_sets)
three_class_performance_dictionary = CurvePlotBuilder.make_three_class_roc(
classifier_summary, '_mesa')
classifier_summary.performance_dictionary = three_class_performance_dictionary
three_class_performance_summaries.append(classifier_summary)
CurvePlotBuilder.combine_sw_and_three_class_plots(
attributed_classifier, 1, 'mesa')
TableBuilder.print_table_three_class(three_class_performance_summaries)
CurvePlotBuilder.combine_plots_as_grid(classifiers, 1,
'_mesa_three_class_roc')
def figures_leave_one_out_three_class_performance():
attributed_classifier = AttributedClassifier(
name='Neural Net',
classifier=MLPClassifier(activation='relu',
hidden_layer_sizes=(15, 15, 15),
max_iter=1000,
alpha=0.01,
solver='lbfgs'))
feature_sets = utils.get_base_feature_sets()
if Constants.VERBOSE:
print('Running ' + attributed_classifier.name + '...')
classifier_summary = ThreeClassClassifierSummaryBuilder.build_leave_one_out(
attributed_classifier, feature_sets)
PerformancePlotBuilder.make_bland_altman(classifier_summary)
def figures_leave_one_out_sleep_wake_performance():
attributed_classifier = AttributedClassifier(
name='Neural Net',
classifier=MLPClassifier(activation='relu',
hidden_layer_sizes=(15, 15, 15),
max_iter=1000,
alpha=0.01,
solver='lbfgs'))
feature_sets = [[
FeatureType.count, FeatureType.heart_rate, FeatureType.circadian_model
]]
if Constants.VERBOSE:
print('Running ' + attributed_classifier.name + '...')
classifier_summary = SleepWakeClassifierSummaryBuilder.build_leave_one_out(
attributed_classifier, feature_sets)
PerformancePlotBuilder.make_histogram_with_thresholds(classifier_summary)
PerformancePlotBuilder.make_single_threshold_histograms(classifier_summary)
def test_make_histogram_with_thresholds(self, mock_plt, mock_np,
mock_performance_summarizer,
mock_image, mock_image_draw,
mock_image_font):
raw_performances = [
'raw_performance_placeholder_1', 'raw_performance_placeholder_2'
]
performance_dictionary = {tuple([FeatureType.count]): raw_performances}
attributed_classifier = AttributedClassifier(
name="Logistic Regression", classifier=LogisticRegression())
classifier_summary = ClassifierSummary(
attributed_classifier=attributed_classifier,
performance_dictionary=performance_dictionary)
mock_ax = [[MagicMock(), MagicMock()], [MagicMock(),
MagicMock()],
[MagicMock(), MagicMock()], [MagicMock(),
MagicMock()]]
number_of_subjects = 2
mock_plt.subplots.return_value = 'placeholder1', mock_ax
mock_np.zeros.side_effect = [
np.zeros((number_of_subjects, 4)),
np.zeros((number_of_subjects, 4))
]
dt = 0.02
expected_range = np.arange(0, 1 + dt, dt)
mock_np.arange.return_value = expected_range
mock_subject1_threshold1 = MagicMock()
mock_subject1_threshold2 = MagicMock()
mock_subject1_threshold3 = MagicMock()
mock_subject1_threshold4 = MagicMock()
mock_subject2_threshold1 = MagicMock()
mock_subject2_threshold2 = MagicMock()
mock_subject2_threshold3 = MagicMock()
mock_subject2_threshold4 = MagicMock()
mock_subject1_threshold1.accuracy = 0
mock_subject1_threshold1.wake_correct = 1
mock_subject1_threshold2.accuracy = 2
mock_subject1_threshold2.wake_correct = 3
mock_subject1_threshold3.accuracy = 4
mock_subject1_threshold3.wake_correct = 5
mock_subject1_threshold4.accuracy = 6
mock_subject1_threshold4.wake_correct = 7
mock_subject2_threshold1.accuracy = 10
mock_subject2_threshold1.wake_correct = 11
mock_subject2_threshold2.accuracy = 12
mock_subject2_threshold2.wake_correct = 13
mock_subject2_threshold3.accuracy = 14
mock_subject2_threshold3.wake_correct = 15
mock_subject2_threshold4.accuracy = 16
mock_subject2_threshold4.wake_correct = 17
mock_performance_summarizer.summarize_thresholds.side_effect = [
([1, 2, 3, 4], [
mock_subject1_threshold1, mock_subject1_threshold2,
mock_subject1_threshold3, mock_subject1_threshold4
]),
([1, 2, 3, 4], [
mock_subject2_threshold1, mock_subject2_threshold2,
mock_subject2_threshold3, mock_subject2_threshold4
])
]
file_save_name = str(
Constants.FIGURE_FILE_PATH
) + '/' + 'Motion only_Logistic Regression_histograms_with_thresholds.png'
mock_image.open.return_value = mock_opened_image = MagicMock()
mock_opened_image.size = 100, 200
mock_image.new.return_value = new_image = MagicMock()
mock_image_draw.Draw.return_value = image_draw = MagicMock()
mock_image_font.truetype.return_value = font = "true type font"
PerformancePlotBuilder.make_histogram_with_thresholds(
classifier_summary)
mock_plt.subplots.assert_called_once_with(nrows=4,
ncols=2,
figsize=(8, 8),
sharex=True,
sharey=True)
mock_np.zeros.assert_has_calls([call((2, 4)), call((2, 4))])
mock_performance_summarizer.summarize_thresholds.assert_has_calls(
[call([raw_performances[0]]),
call([raw_performances[1]])])
mock_ax[0][0].hist.assert_called_once_with([0, 10],
bins=expected_range,
color="skyblue",
ec="skyblue")
mock_ax[0][1].hist.assert_called_once_with([1, 11],
bins=expected_range,
color="lightsalmon",
ec="lightsalmon")
mock_ax[1][0].hist.assert_called_once_with([2, 12],
bins=expected_range,
color="skyblue",
ec="skyblue")
mock_ax[1][1].hist.assert_called_once_with([3, 13],
bins=expected_range,
color="lightsalmon",
ec="lightsalmon")
mock_ax[2][0].hist.assert_called_once_with([4, 14],
bins=expected_range,
color="skyblue",
ec="skyblue")
mock_ax[2][1].hist.assert_called_once_with([5, 15],
bins=expected_range,
color="lightsalmon",
ec="lightsalmon")
mock_ax[3][0].hist.assert_called_once_with([6, 16],
bins=expected_range,
color="skyblue",
ec="skyblue")
mock_ax[3][1].hist.assert_called_once_with([7, 17],
bins=expected_range,
color="lightsalmon",
ec="lightsalmon")
mock_ax[3][0].set_xlabel.assert_called_once_with('Accuracy',
fontsize=16,
fontname='Arial')
mock_ax[3][1].set_xlabel.assert_called_once_with('Wake correct',
fontsize=16,
fontname='Arial')
mock_ax[3][0].set_ylabel.assert_called_once_with('Count',
fontsize=16,
fontname='Arial')
mock_ax[3][0].set_xlim.assert_called_once_with((0, 1))
mock_ax[3][1].set_xlim.assert_called_once_with((0, 1))
mock_plt.tight_layout.assert_called_once_with()
mock_plt.savefig.assert_called_once_with(file_save_name, dpi=300)
mock_plt.close.assert_called_once_with()
mock_image.open.assert_called_once_with(file_save_name)
mock_image.new.assert_called_once_with('RGB', (int(
(1 + 0.3) * 100), 200), "white")
new_image.paste.assert_called_once_with(mock_opened_image,
(int(0.3 * 100), 0))
mock_image_draw.Draw.assert_called_once_with(new_image)
mock_image_font.truetype.assert_called_once_with(
'/Library/Fonts/Arial Unicode.ttf', 75)
image_draw.text.assert_has_calls([
call((int(0.3 * 100 / 3), int((200 * 0.9) * 0.125)),
"TPR = 0.8", (0, 0, 0),
font=font),
call((int(0.3 * 100 / 3), int((200 * 0.9) * 0.375)),
"TPR = 0.9", (0, 0, 0),
font=font),
call((int(0.3 * 100 / 3), int((200 * 0.9) * 0.625)),
"TPR = 0.93", (0, 0, 0),
font=font),
call((int(0.3 * 100 / 3), int((200 * 0.9) * 0.875)),
"TPR = 0.95", (0, 0, 0),
font=font)
])
new_image.save.assert_called_once_with(
str(Constants.FIGURE_FILE_PATH) + '/' +
'figure_threshold_histogram.png')