def main():
# Set up file names and locations.
DATA_PATH = Path('./intermediate_datafiles/')
DATASET_FNAME = sys.argv[1] if len(sys.argv) > 1 else 'chapter2_result.csv'
RESULT_FNAME = sys.argv[2] if len(
sys.argv) > 2 else 'chapter3_result_outliers.csv'
# Next, import the data from the specified location and parse the date index.
try:
dataset = pd.read_csv(Path(DATA_PATH / DATASET_FNAME), index_col=0)
dataset.index = pd.to_datetime(dataset.index)
except IOError as e:
print(
'File not found, try to run the preceding crowdsignals scripts first!'
)
raise e
# We'll create an instance of our visualization class to plot the results.
DataViz = VisualizeDataset(__file__)
# Compute the number of milliseconds covered by an instance using the first two rows.
milliseconds_per_instance = (dataset.index[1] -
dataset.index[0]).microseconds / 1000
# Step 1: Let us see whether we have some outliers we would prefer to remove.
# Determine the columns we want to experiment on.
outlier_columns = ['acc_phone_x', 'light_phone_lux']
# Create the outlier classes.
OutlierDistr = DistributionBasedOutlierDetection()
OutlierDist = DistanceBasedOutlierDetection()
# And investigate the approaches for all relevant attributes.
for col in outlier_columns:
print(f"Applying outlier criteria for column {col}")
# And try out all different approaches. Note that we have done some optimization
# of the parameter values for each of the approaches by visual inspection.
dataset = OutlierDistr.chauvenet(dataset, col)
DataViz.plot_binary_outliers(dataset, col, col + '_outlier')
dataset = OutlierDistr.mixture_model(dataset, col)
DataViz.plot_dataset(dataset, [col, col + '_mixture'],
['exact', 'exact'], ['line', 'points'])
# This requires:
# n_data_points * n_data_points * point_size =
# 31839 * 31839 * 32 bits = ~4GB available memory
try:
dataset = OutlierDist.simple_distance_based(
dataset, [col], 'euclidean', 0.10, 0.99)
DataViz.plot_binary_outliers(dataset, col, 'simple_dist_outlier')
except MemoryError as e:
print(
'Not enough memory available for simple distance-based outlier detection...'
)
print('Skipping.')
try:
dataset = OutlierDist.local_outlier_factor(dataset, [col],
'euclidean', 5)
DataViz.plot_dataset(dataset, [col, 'lof'], ['exact', 'exact'],
['line', 'points'])
except MemoryError as e:
print('Not enough memory available for lof...')
print('Skipping.')
# Remove all the stuff from the dataset again.
cols_to_remove = [
col + '_outlier', col + '_mixture', 'simple_dist_outlier', 'lof'
]
for to_remove in cols_to_remove:
if to_remove in dataset:
del dataset[to_remove]
# We take Chauvenet's criterion and apply it to all but the label data...
for col in [c for c in dataset.columns if not 'label' in c]:
print(f'Measurement is now: {col}')
dataset = OutlierDistr.chauvenet(dataset, col)
dataset.loc[dataset[f'{col}_outlier'] == True, col] = np.nan
del dataset[col + '_outlier']
dataset.to_csv(DATA_PATH / RESULT_FNAME)
def main():
# Set up file names and locations.
# Next, import the data from the specified location and parse the date index.
dataset = pickle.load(open('datasets\dataframes\df_concat_with_labels.pkl', 'rb'))
# We'll create an instance of our visualization class to plot the results.
DataViz = VisualizeDataset(__file__, show=False)
# Compute the number of milliseconds covered by an instance using the first two rows.
milliseconds_per_instance = (dataset.index[1] - dataset.index[0]).microseconds/1000
# Step 1: Let us see whether we have some outliers we would prefer to remove.
# Determine the columns we want to experiment on.
outlier_columns = [
"Acceleration x (m/s^2)","Acceleration y (m/s^2)","Acceleration z (m/s^2)",
"Magnetic field x (µT)","Magnetic field y (µT)","Magnetic field z (µT)",
"Gyroscope x (rad/s)","Gyroscope y (rad/s)","Gyroscope z (rad/s)",
"Linear Acceleration x (m/s^2)","Linear Acceleration y (m/s^2)","Linear Acceleration z (m/s^2)",
]
print(dataset.columns)
# Create the outlier classes.
OutlierDistr = DistributionBasedOutlierDetection()
OutlierDist = DistanceBasedOutlierDetection()
# And investigate the approaches for all relevant attributes.
for col in outlier_columns:
print(f"Applying outlier criteria for column {col}")
# And try out all different approaches. Note that we have done some optimization
# of the parameter values for each of the approaches by visual inspection.
dataset = OutlierDistr.chauvenet(dataset, col)
print(dataset.shape)
# DataViz.plot_binary_outliers(dataset, col, col + '_outlier')
dataset = OutlierDistr.mixture_model(dataset, col)
print(dataset.shape)
# DataViz.plot_dataset(dataset, [col, col + '_mixture'], ['exact','exact'], ['line', 'points'])
# This requires:
# n_data_points * n_data_points * point_size =
# 31839 * 31839 * 32 bits = ~4GB available memory
try:
dataset = OutlierDist.simple_distance_based(dataset, [col], 'euclidean', 0.10, 0.99)
# DataViz.plot_binary_outliers(dataset, col, 'simple_dist_outlier')
print(dataset['simple_dist_outlier'].mean())
except MemoryError as e:
print('Not enough memory available for simple distance-based outlier detection...')
print('Skipping.')
try:
dataset = OutlierDist.local_outlier_factor(dataset, [col], 'euclidean', 5)
# DataViz.plot_dataset(dataset, [col, 'lof'], ['exact','exact'], ['line', 'points'])
except MemoryError as e:
print('Not enough memory available for lof...')
print('Skipping.')
# Remove all the stuff from the dataset again.
cols_to_remove = [col + '_outlier', col + '_mixture', 'simple_dist_outlier', 'lof']
for to_remove in cols_to_remove:
if to_remove in dataset:
del dataset[to_remove]
# We take Chauvenet's criterion and apply it to all but the label data...
for col in [c for c in dataset.columns if not 'label' in c]:
print(f'Measurement is now: {col}')
dataset = OutlierDistr.chauvenet(dataset, col)
dataset.loc[dataset[f'{col}_outlier'] == True, col] = np.nan
del dataset[col + '_outlier']
pickle.dump(dataset, open('concat_outliers.pkl', 'wb'))
def main():
# Import the data from the specified location and parse the date index
try:
dataset = pd.read_csv(Path(DATA_PATH / DATASET_FILENAME), index_col=0)
dataset.index = pd.to_datetime(dataset.index)
except IOError as e:
print('File not found, try to run the preceding crowdsignals scripts first!')
raise e
# Create an instance of visualization class to plot the results
DataViz = VisualizeDataset(module_path=__file__)
# Create the outlier classes
OutlierDistribution = DistributionBasedOutlierDetection()
OutlierDistance = DistanceBasedOutlierDetection()
# Step 1: If requested, see whether there are some outliers that need to be preferably removed
# Set the columns to experiment on
outlier_columns = ['acc_phone_x', 'light_phone_lux']
if FLAGS.mode == 'chauvenet':
# Investigate the outlier columns using chauvenet criterium
for col in outlier_columns:
print(f"Applying chauvenet outlier criteria for column {col}")
dataset = OutlierDistribution.chauvenet(data_table=dataset, col=col)
DataViz.plot_binary_outliers(data_table=dataset, col=col, outlier_col=f'{col}_outlier')
elif FLAGS.mode == 'mixture':
# Investigate the outlier columns using mixture models
for col in outlier_columns:
print(f"Applying mixture model for column {col}")
dataset = OutlierDistribution.mixture_model(data_table=dataset, col=col, components=3)
DataViz.plot_dataset(data_table=dataset, columns=[col, f'{col}_mixture'], match=['exact', 'exact'],
display=['line', 'points'])
elif FLAGS.mode == 'distance':
for col in outlier_columns:
print(f"Applying distance based outlier detection for column {col}")
# This step requires:
# n_data_points * n_data_points * point_size = 31839 * 31839 * 32 bits = ~4GB available memory
try:
dataset = OutlierDistance.simple_distance_based(data_table=dataset, cols=[col], d_function='euclidean',
d_min=FLAGS.dmin, f_min=FLAGS.fmin)
DataViz.plot_binary_outliers(data_table=dataset, col=col, outlier_col='simple_dist_outlier')
except MemoryError:
print('Not enough memory available for simple distance-based outlier detection...')
print('Skipping.')
elif FLAGS.mode == 'LOF':
for col in outlier_columns:
print(f"Applying Local outlier factor for column {col}")
try:
dataset = OutlierDistance.local_outlier_factor(data_table=dataset, cols=[col], d_function='euclidean',
k=FLAGS.K)
DataViz.plot_dataset(data_table=dataset, columns=[col, 'lof'], match=['exact', 'exact'],
display=['line', 'points'])
except MemoryError:
print('Not enough memory available for local outlier factor...')
print('Skipping.')
elif FLAGS.mode == 'final':
# Take Chauvenet's criterion and apply it to all but the label column in the main dataset
for col in [c for c in dataset.columns if 'label' not in c]:
print(f'Measurement is now: {col}')
dataset = OutlierDistribution.chauvenet(data_table=dataset, col=col)
dataset.loc[dataset[f'{col}_outlier'], col] = np.nan
del dataset[col + '_outlier']
dataset.to_csv(DATA_PATH / RESULT_FILENAME)
class OutlierExperiment:
def __init__(self, data_path, data_file):
self.dataset = pd.read_csv(Path(data_path / data_file), index_col=0)
self.dataset = self.dataset
self.dataset.index = pd.to_datetime(self.dataset.index)
self.DataViz = VisualizeDataset(__file__, show=False)
self.outlier_columns = ['acc_phone_x', 'light_phone_lux']
self.OutlierDistr = DistributionBasedOutlierDetection()
self.OutlierDist = DistanceBasedOutlierDetection()
self.original_columns = self.dataset.columns
self.num_outliers = {'acc_phone_x': 0, 'light_phone_lux': 0}
def remove_columns(self):
for to_remove in self.dataset.columns:
if to_remove not in self.original_columns:
del self.dataset[to_remove]
def chauvenet(self, C):
original_columns = self.dataset.columns
for col in self.outlier_columns:
print(f"Applying outlier criteria for column {col}")
self.dataset = self.OutlierDistr.chauvenet(self.dataset, col, C)
self.DataViz.plot_binary_outliers(self.dataset, col,
col + '_outlier')
self.num_outliers[col] = self.dataset[self.dataset[
col + '_outlier'] == 1][col].size / self.dataset[col].size
def mixture_model(self, n):
for col in self.outlier_columns:
print(f"Applying outlier criteria for column {col}")
self.dataset = self.OutlierDistr.mixture_model(
self.dataset, col, n)
self.DataViz.plot_dataset(self.dataset, [col, col + '_mixture'],
['exact', 'exact'], ['line', 'points'])
self.num_outliers[col] = self.dataset[
col + '_mixture'].sum() / self.dataset[col + '_mixture'].size
print(self.dataset[col + '_mixture'].max())
if self.num_outliers[col] > 1:
print(self.dataset[col + '_mixture'])
def simple_distance_based(self, d_min, f_min):
for col in self.outlier_columns:
print(f"Applying outlier criteria for column {col}")
self.dataset = self.OutlierDist.simple_distance_based(
self.dataset, [col], 'euclidean', d_min, f_min)
self.DataViz.plot_binary_outliers(self.dataset, col,
'simple_dist_outlier')
self.num_outliers[col] = self.dataset[
self.dataset['simple_dist_outlier'] ==
1][col].size / self.dataset[col].size
self.remove_columns()
def local_outlier_factor(self, k):
for col in self.outlier_columns:
print(f"Applying outlier criteria for column {col}")
self.dataset = self.OutlierDist.local_outlier_factor(
self.dataset, [col], 'euclidean', k)
self.DataViz.plot_dataset(self.dataset, [col, 'lof'],
['exact', 'exact'], ['line', 'points'])
self.num_outliers[col] = self.dataset[
self.dataset['lof'] == 1][col].size / self.dataset[col].size
self.remove_columns()
# n_data_points * n_data_points * point_size =
# 31839 * 31839 * 64 bits = ~8GB available memory
try:
dataset = OutlierDist.simple_distance_based(dataset, [col],
'euclidean', 0.10, 0.99)
DataViz.plot_binary_outliers(dataset, col, 'simple_dist_outlier')
except MemoryError as e:
print(
'Not enough memory available for simple distance-based outlier detection...'
)
print('Skipping.')
try:
dataset = OutlierDist.local_outlier_factor(dataset, [col], 'euclidean',
5)
DataViz.plot_dataset(dataset, [col, 'lof'], ['exact', 'exact'],
['line', 'points'])
except MemoryError as e:
print('Not enough memory available for lof...')
print('Skipping.')
# Remove all the stuff from the dataset again.
cols_to_remove = [
col + '_outlier', col + '_mixture', 'simple_dist_outlier', 'lof'
]
for to_remove in cols_to_remove:
if to_remove in dataset:
del dataset[to_remove]
# We take Chauvenet's criterion and apply it to all but the label data...
['exact', 'exact'], ['line', 'points'])
# This requires:
# n_data_points * n_data_points * point_size =
# 31839 * 31839 * 64 bits = ~8GB available memory
try:
dataset_cs = OutlierDist.simple_distance_based(dataset_cs, [col],
'euclidean', dmin, fmin)
DataViz.plot_binary_outliers(dataset_cs, col, 'simple_dist_outlier')
except MemoryError as e:
print(
'Not enough memory available for simple distance-based outlier detection...'
)
print('Skipping.')
try:
dataset_cs = OutlierDist.local_outlier_factor(dataset_cs, [col],
'euclidean', k)
DataViz.plot_dataset(dataset_cs, [col, 'lof'], ['exact', 'exact'],
['line', 'points'])
except MemoryError as e:
print('Not enough memory available for lof...')
print('Skipping.')
# Remove all the stuff from the dataset again.
cols_to_remove = [
col + '_outlier', col + '_mixture', 'simple_dist_outlier', 'lof'
]
for to_remove in cols_to_remove:
if to_remove in dataset_cs:
del dataset_cs[to_remove]
# We take Chauvent's criterion and apply it to all but the label data...