def fetch_tuebingen_2019(keep_original=True, cache=True, retain_corrections=False):
"""
Fetches the tuebingen_2019 dataset from the internet. The original dataset or its cached version
is stored in the :ref:`data home <storage>` folder.
Parameters
----------
keep_original : bool, default=True
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool, default=True
Determines whether the data object should be stored as a binary file for faster
succeeding access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
retain_corrections : bool, optional, default=False
When set to *true* data points that are changed or dropped during preprocessing
are listed in the respective attributes of the data object. Fore more information
about the attributes refer to the :ref:`user guide <error_correction>`.
Returns
-------
data : object
"""
dataset_name = 'tuebingen_2019'
def load_tuebingen_2019(folder_path):
return act_assist.load(folder_path, subjects=['M'])
data = _fetch_handler(keep_original, cache, dataset_name,
TUE_2019_FILENAME, TUE_2019_URL,
load_tuebingen_2019)
return data
def fetch_casas_aruba(keep_original=True,
cache=True,
retain_corrections=False):
"""
Fetches the casas aruba dataset from the internet. The original dataset or its cached version
is stored in the :ref:`data home <storage>` folder.
Parameters
----------
keep_original : bool, default=True
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool, default=True
Determines whether the data object should be stored as a binary file for quicker access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
retain_corrections : bool, default=False
When set to *true* data points that are changed or dropped during preprocessing
are listed in the respective attributes of the data object. Fore more information
about the attributes refer to the :ref:`user guide <error_correction>`.
Examples
--------
>>> from pyadlml.dataset import fetch_casas_aruba
>>> data = fetch_casas_aruba()
>>> dir(data)
>>> [..., df_activities, df_devices, ...]
Returns
-------
data : object
"""
dataset_name = 'casas_aruba'
def load_casas_aruba(folder_path):
_fix_data(os.path.join(folder_path, "data"))
data_path = os.path.join(folder_path, 'corrected_data.csv')
df = _load_df(data_path)
df_dev = _get_devices_df(df)
df_act = _get_activity_df(df)
df_dev = correct_devices(df_dev)
df_act, cor_lst = correct_activities(df_act)
lst_act = df_act[ACTIVITY].unique()
lst_dev = df_dev[DEVICE].unique()
data = Data(df_act, df_dev, activity_list=lst_act, device_list=lst_dev)
return data
data = _fetch_handler(keep_original, cache, dataset_name,
CASAS_ARUBA_FILENAME, CASAS_ARUBA_URL,
load_casas_aruba)
return data
def fetch_amsterdam(keep_original=False, cache=True, retain_corrections=False):
"""
Fetches the amsterdam dataset from the internet. The original dataset or its cached version
is stored in the :ref:`data home <storage>` folder.
Parameters
----------
keep_original : bool, default=False
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool, default=True
Determines whether the data object should be stored as a binary file for quicker access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
retain_corrections : bool, default=False
When set to *true*, data points that change or drop during preprocessing
are listed in respective attributes of the data object. Fore more information
about error correction refer to the :ref:`user guide <error_correction>`.
Examples
--------
>>> from pyadlml.dataset import fetch_amsterdam
>>> data = fetch_amsterdam()
>>> dir(data)
>>> [..., df_activities, df_devices, ...]
Returns
-------
data : object
"""
dataset_name = 'amsterdam'
def load_amsterdam(folder_path):
device_fp = os.path.join(folder_path, "kasterenSenseData.txt")
activity_fp = os.path.join(folder_path, "kasterenActData.txt")
df_act = _load_activities(activity_fp)
df_dev = _load_devices(device_fp)
df_act, cor_lst = correct_activities(df_act)
df_dev = correct_devices(df_dev)
lst_act = df_act[ACTIVITY].unique()
lst_dev = df_dev[DEVICE].unique()
data = Data(df_act, df_dev, activity_list=lst_act, device_list=lst_dev)
if retain_corrections:
data.correction_activities = cor_lst
return data
data = _fetch_handler(keep_original, cache, dataset_name,
AMSTERDAM_FILENAME, AMSTERDAM_URL,
load_amsterdam)
return data
def fetch_uci_adl_binary(keep_original=True,
cache=True,
retain_corrections=False,
subject='OrdonezA'):
"""
Parameters
----------
keep_original : bool, default=True
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool
Determines whether the data object should be stored as a binary file for quicker access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
retain_corrections : bool
When set to *true* data points that are changed or dropped during preprocessing
are listed in the respective attributes of the data object. Fore more information
about the attributes refer to the :ref:`user guide <error_correction>`.
subject : str of {'OrdonezA', 'OrdonezB'}, default='OrdonezA'
decides which dataset of the two houses is loaded.
Returns
-------
data : object
"""
assert subject in ['OrdonezA', 'OrdonezB']
dataset_name = 'uci_adl_binary'
def load_uci_adl_binary(folder_path):
sub_dev_file = os.path.join(folder_path,
'{}_Sensors.txt'.format(subject))
if subject == 'OrdonezB':
fix_OrdonezB_ADLS(os.path.join(folder_path, 'OrdonezB_ADLs.txt'))
sub_act_file = os.path.join(folder_path,
'{}_ADLs_corr.txt'.format(subject))
else:
sub_act_file = os.path.join(folder_path,
'{}_ADLs.txt'.format(subject))
return load(sub_dev_file, sub_act_file, retain_corrections, subject)
data = _fetch_handler(keep_original,
cache,
dataset_name,
UCI_ADL_BINARY_FILENAME,
UCI_ADL_BINARY_URL,
load_uci_adl_binary,
data_postfix=subject)
return data
def fetch_amsterdam(keep_original=True,
cache=True,
remember_corrections=False):
"""
Fetches the amsterdam dataset from the internet. The original dataset or its cached version
is stored in the :ref:`data home <storage>` folder.
Parameters
----------
keep_original : bool, default=True
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool, default=True
Determines whether the data object should be stored as a binary file for quicker access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
remember_corrections : bool
When set to *true* data points that are changed or dropped during preprocessing
are listed in the respective attributes of the data object. Fore more information
about the attributes refer to the :ref:`user guide <error_correction>`.
Examples
--------
>>> from pyadlml.dataset import fetch_amsterdam
>>> data = fetch_amsterdam()
>>> dir(data)
>>> [..., df_activities, df_devices, ...]
Returns
-------
data : object
"""
dataset_name = 'amsterdam'
def load_amsterdam(folder_path):
sensorData = os.path.join(folder_path, "kasterenSenseData.txt")
activityData = os.path.join(folder_path, "kasterenActData.txt")
return amsterdam.load(sensorData, activityData)
data = _fetch_handler(keep_original, cache, dataset_name,
AMSTERDAM_FILENAME, AMSTERDAM_URL, load_amsterdam)
return data
def fetch_mitlab(keep_original=True, cache=True, subject='subject1'):
"""
Parameters
----------
keep_original : bool, default=True
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool, default=True
Determines whether the data object should be stored as a binary file for quicker access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
remember_corrections : bool
When set to *true* data points that are changed or dropped during preprocessing
are listed in the respective attributes of the data object. Fore more information
about the attribtues refer to the :ref:`user guide <error_correction>`.
Returns
-------
data : object
"""
assert subject in ['subject1', 'subject2']
dataset_name = 'mitlab'
def load_mitlab(folder_path):
sub_act = os.path.join(folder_path, subject, "Activities.csv")
sub_dev = os.path.join(folder_path, subject, "sensors.csv")
sub_data = os.path.join(folder_path, subject, "activities_data.csv")
return mitlab.load(sub_dev, sub_act, sub_data)
data = _fetch_handler(keep_original,
cache,
dataset_name,
MITLAB_FILENAME,
MITLAB_URL,
load_mitlab,
data_postfix=subject)
return data
def fetch_aras(keep_original=True, cache=True, retain_corrections=False):
"""
Fetches the aras dataset from the internet. The original dataset or its cached version
is stored in the :ref:`data home <storage>` folder.
Parameters
----------
keep_original : bool, default=True
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool, default=True
Determines whether the data object should be stored as a binary file for quicker access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
retain_corrections : bool, default=False
When set to *true* data points that are changed or dropped during preprocessing
are listed in the respective attributes of the data object. Fore more information
about the attributes refer to the :ref:`user guide <error_correction>`.
Examples
--------
>>> from pyadlml.dataset import fetch_aras
>>> data = fetch_aras()
>>> dir(data)
>>> [..., df_activities, df_devices, ...]
Returns
-------
data : object
"""
dataset_name = 'aras'
def load_aras(data_path):
device_map = _get_device_map(data_path)
activity_map = _get_activity_map(data_path)
df = _read_data(data_path, activity_map, device_map)
df_res1_act = _create_activity_df(df, 'Resident 1')
df_res2_act = _create_activity_df(df, 'Resident 2')
# TODO correct activities
assert not _is_activity_overlapping(df_res1_act) \
or not _is_activity_overlapping(df_res2_act)
df_dev = _create_device_df(df)
df_dev = correct_devices(df_dev)
lst_res1_act = df_res1_act[ACTIVITY].unique()
lst_res2_act = df_res2_act[ACTIVITY].unique()
lst_dev = df_dev[DEVICE].unique()
data = Data(None, df_dev, activity_list=None, device_list=lst_dev)
data.df_dev_map = device_map
data.df_act_map = activity_map
data.df_activities_res1 = df_res1_act
data.lst_activities_res1 = lst_res1_act
data.df_activities_res2 = df_res2_act
data.lst_activities_res2 = lst_res2_act
if retain_corrections:
data.correction_activities_res1 = []
data.correction_activities_res2 = []
return data
data = _fetch_handler(keep_original, cache, dataset_name, ARAS_FILENAME,
ARAS_URL, load_aras)
return data
def fetch_mitlab(keep_original=False,
cache=True,
retain_corrections=False,
subject='subject1'):
"""
Fetches the :ref:`mitlab <ds_mitlab>` dataset from the internet. The original dataset or its cached version
is stored in the :ref:`data home <storage>` folder.
Parameters
----------
keep_original : bool, default=True
Determines whether the original dataset is deleted after downloading
or kept on the hard drive.
cache : bool, default=True
Determines whether the data object should be stored as a binary file for quicker access.
For more information how caching is used refer to the :ref:`user guide <storage>`.
retain_corrections : bool, default=False
When set to *true*, data points that change or drop during preprocessing
are listed in respective attributes of the data object. Fore more information
about error correction refer to the :ref:`user guide <error_correction>`.
subject : str of {'subject1', 'subject2'}
determines
Returns
-------
data : object
"""
assert subject in ['subject1', 'subject2']
dataset_name = 'mitlab'
def load_mitlab(folder_path):
act_path = os.path.join(folder_path, subject, "Activities.csv")
dev_path = os.path.join(folder_path, subject, "sensors.csv")
data_path = os.path.join(folder_path, subject, "activities_data.csv")
df_dev_map = _load_device_map(dev_path)
df_act_map = _load_activity_map(act_path)
df_dev, df_act = _read_data(data_path, df_dev_map, df_act_map)
df_act, cor_lst = correct_activities(df_act)
df_dev = correct_devices(df_dev)
lst_act = df_act[ACTIVITY].unique()
lst_dev = df_dev[DEVICE].unique()
data = Data(df_act, df_dev, activity_list=lst_act, device_list=lst_dev)
data.df_dev_map = df_dev_map
data.df_act_map = df_act_map
if retain_corrections:
data.correction_activities = cor_lst
return data
data = _fetch_handler(keep_original,
cache,
dataset_name,
MITLAB_FILENAME,
MITLAB_URL,
load_mitlab,
data_postfix=subject)
return data