def compute_triaxial_norm(df, remove_triaxial_vectors=True):
columns = df.columns
sensors = [
x[:-2] for x in columns
if ('_x' in x) and ((x[:-2] + '_y') in columns) and ((x[:-2] +
'_z') in columns)
]
for sens in sensors:
cols = df.loc[:, [
"{}_x".format(sens), "{}_y".format(sens), "{}_z".format(sens)
]].values
norms = np.linalg.norm(cols, axis=1)
if remove_triaxial_vectors:
df.drop(labels=mul_str_arr([sens], ["x", "y", "z"]),
axis=1,
inplace=True)
df.insert(loc=len(df.columns),
column="{}_magnitude".format(sens),
value=norms)
return df
# "label",
# ]
samples_table = "__".join([DATASET_NAME, "samples"])
samples_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
("timestamp", "DECIMAL"),
("subject_id", "INTEGER"),
("activity_id", "INTEGER"),
])
samples_n_columns = len(samples_schema)
sensor_readings_table = "__".join([DATASET_NAME, "sensor_readings"])
sensor_readings_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
] + [(k, "FLOAT") for k in mul_str_arr(["chest_acc"], ["x", "y", "z"]) +
["ecg_1", "ecg_2"] + mul_str_arr(["left_ankle", "right_lower_arm"],
["acc", "gyro", "magn"], ["x", "y", "z"])
] + [("FOREIGN KEY",
"(sample_id) REFERENCES {}(sample_id)".format(samples_table))])
sensor_readings_n_columns = len(sensor_readings_schema) - 1
def _load_dataset(path, loader):
return OrderedDict(
(int(filepath[15:-4]), loader(os.path.join(path, DATA_PATH, filepath)))
for filepath in os.listdir(os.path.join(path, DATA_PATH))
if filepath.endswith('.log'))
# insert timestamp and subject id
# tbls_new = []
# for subject in range(len(tbls)):
samples_table = "__".join([DATASET_NAME, "samples"])
samples_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
("timestamp", "DECIMAL"),
("subject_id", "INTEGER"),
("activity_id", "INTEGER"),
])
samples_n_columns = len(samples_schema)
sensor_readings_table = "__".join([DATASET_NAME, "sensor_readings"])
sensor_readings_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
("heart_rate", "FLOAT"),
] + [
(k, "FLOAT")
for k in mul_str_arr(["hand", "chest", "ankle"], ["temperature"] +
mul_str_arr(["acc", "gyro", "magn"], ["x", "y", "z"]))
] + [("FOREIGN KEY",
"(sample_id) REFERENCES {}(sample_id)".format(samples_table))])
sensor_readings_n_columns = len(sensor_readings_schema) - 1
# column_headings = [
# "timestamp",
# "activity_id",
# "heart_rate",
# ]
def _load_dataset(path, loader_func):
_data_path = os.path.join(path, DATA_PATH)
tbls = []
subdirectories = ["Protocol", "Optional"]
# "label",
# ]
samples_table = "__".join([DATASET_NAME, "samples"])
samples_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
("timestamp", "DECIMAL"),
("subject_id", "INTEGER"),
("activity_id", "INTEGER"),
])
samples_n_columns = len(samples_schema)
sensor_readings_table = "__".join([DATASET_NAME, "sensor_readings"])
sensor_readings_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
] + [(k, "FLOAT") for k in mul_str_arr(["chest_acc"], ["x", "y", "z"])] + [(
"FOREIGN KEY",
"(sample_id) REFERENCES {}(sample_id)".format(samples_table))])
sensor_readings_n_columns = len(sensor_readings_schema) - 1
def _load_dataset(path, loader):
return OrderedDict(
(int(filepath[:-4]), loader(os.path.join(path, DATA_PATH, filepath)))
for filepath in os.listdir(os.path.join(path, DATA_PATH))
if filepath.endswith('.csv'))
def load_dataset_to_mem(path):
return _load_dataset(path, np.genfromtxt)
# acc unit: m/s^2
# gyro unit: deg/s
distinct_subject_query = """
SELECT DISTINCT subject_id FROM {};
""".format(samples_table)
distinct_activity_query = """
SELECT DISTINCT activity_id FROM {};
""".format(samples_table)
raw_table_valid_data_query = ("""
SELECT
activity_id, subject_id,
""" + ", ".join(
mul_str_arr(["chest_acc"], ["x", "y", "z"]) + ["ecg_1", "ecg_2"] +
mul_str_arr(["left_ankle", "right_lower_arm"], ["acc", "gyro", "magn"],
["x", "y", "z"])) + """
FROM
{0}, {1}
WHERE {0}.sample_id = {1}.sample_id AND activity_id != 0;
""").format(samples_table, sensor_readings_table)
raw_table_valid_data_query_with_subject_id = ("""
SELECT
activity_id, timestamp, subject_id,
""" + ", ".join(
mul_str_arr(["chest_acc"], ["x", "y", "z"]) + ["ecg_1", "ecg_2"] +
mul_str_arr(["left_ankle", "right_lower_arm"], ["acc", "gyro", "magn"],
["x", "y", "z"])) + """
FROM
sampling_interval = Decimal('1.0') / sampling_freq
samples_table = "__".join([DATASET_NAME, "samples"])
samples_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
("timestamp", "DECIMAL"),
("subject_id", "INTEGER"),
("activity_id", "INTEGER"),
])
samples_n_columns = len(samples_schema)
sensor_readings_table = "__".join([DATASET_NAME, "sensor_readings"])
sensor_readings_schema = OrderedDict([
("sample_id", "INTEGER PRIMARY KEY"),
] + [(k, "FLOAT")
for k in mul_str_arr(["body_acc", "body_gyro"], ["x", "y", "z"])] + [(
"FOREIGN KEY",
"(sample_id) REFERENCES {}(sample_id)".format(samples_table))])
sensor_readings_n_columns = len(sensor_readings_schema) - 1
def _load_dataset(path, loaders):
return [
loaders(*[
os.path.join(path, DATA_PATH, subdirectory, filepath)
for filepath in [
"subject_{}.txt".format(subdirectory), "y_{}.txt".format(
subdirectory)
] + [
"Inertial Signals/{}_{}_{}.txt".format(s, a, subdirectory)
for s in ["total_acc", "body_gyro"] for a in ["x", "y", "z"]