nilspodlib.datastream.Datastream

class nilspodlib.datastream.Datastream(data: ndarray, sampling_rate: float = 1.0, columns: Optional[Iterable] = None, calibrated_unit: Optional[str] = None, sensor_type: Optional[str] = None)

Object representing a single set of data of one sensor_type.

Usually it is not required to directly interact with the datastream object (besides accessing the data attribute). Most important functionality can/should be used via a dataset or session object to manage multiple datasets at once.

Attributes:

data

The actual data of the sensor_type as np.array. Can have multiple dimensions depending on the sensor_type.

sensor_type

The name of the sensor_type

is_calibrated

If the sensor_type is in a raw format or the expected final output units

is_factory_calibrated

If the datastream was factory calibrated and hence, provided in physical meaningful units. This should be True if the datastream was loaded using the methods to load datasets and sessions provided in this library.

sampling_rate_hz

The sampling rate of the datastream

Methods

cut([start, stop, step, inplace])	Cut the datastream.
data_as_df([index_as_time, include_units])	Export the datastream as pandas DataFrame using self.columns as colummns.
downsample(factor[, inplace])	Downsample the datastreams by a factor.
norm()	Calculate the norm of the data along the last axis.
normalize()	Get the normalized data.

__init__(data: ndarray, sampling_rate: float = 1.0, columns: Optional[Iterable] = None, calibrated_unit: Optional[str] = None, sensor_type: Optional[str] = None)

Get new datastream instance.

Parameters:

data

The actual data to be stored in the datastream. Should be 2D (even if only 1D data is stored). First dimension should be the time axis and second dimension the different data vector entries (e.g. acc_x, acc_y, acc_z).

sampling_rate

The sampling rate of the datastream. Is used for all calculations that require sampling info.

columns

Optional list of names for the data vector entries. Only used to make it easier to understand the content of a datastream.

calibrated_unit

The expected unit of the datastream after calibration.

sensor_type

Type of sensor_type the data is produced from. This allows to automatically get default values for columns and units from nilspodlib.consts.

cut(start: Optional[int] = None, stop: Optional[int] = None, step: Optional[int] = None, inplace: bool = False) → Self

Cut the datastream.

This is equivalent to applying the following slicing to the data of the datastream: array[start:stop:step]

Parameters:

start

Start index

stop

Stop index

step

Step size of the cut

inplace

If True this methods modifies the current datastream object. If False, a copy of the datastream is returned.

data_as_df(index_as_time: bool = True, include_units=False) → DataFrame

Export the datastream as pandas DataFrame using self.columns as colummns.

Parameters:

index_as_time

If True the index will be divided by the sampling rate to represent time since start of the measurement

include_units

If True the column names will have the unit of the datastream concatenated with an

Returns:

DataFrame

downsample(factor: int, inplace: bool = False) → Self

Downsample the datastreams by a factor.

Parameters:

factor

Factor by which the datastream should be downsampled.

inplace

If True this methods modifies the current datastream object. If False, a copy of the datastream is returned.

norm() → ndarray

Calculate the norm of the data along the last axis.

This will provide the vector norm for each time point.

normalize() → Self

Get the normalized data.

Normalization is performed by dividing the data by its maximum absolute value. This will always return a copy of the datastream.

Examples using nilspodlib.datastream.Datastream

Single Dataset

Single Dataset