MedianComputer

class stcal.outlier_detection.median.MedianComputer(full_shape: tuple, in_memory: bool, buffer_size: int | None = None, dtype: str | dtype = 'float32', tempdir: str = '')

Bases: object

Class to efficiently compute a median.

Top-level class to treat median computation uniformly, whether in memory or on disk.

Initialize MedianComputer.

Parameters:

• full_shape – The shape of the full input dataset.

• in_memory – Whether to perform the median computation in memory or using temporary files on disk to save memory.

• buffer_size – The buffer size for the median computation, units of bytes. Has no effect if in_memory is True.

• dtype – The data type of the input data.

• tempdir – The parent directory in which to create the temporary directory. Default is the current working directory.

Methods Summary

append(data[, idx])	Append data to the median computer.
evaluate()	Compute the median data from the input data.

Methods Documentation

append(data: ndarray, idx: int | None = None) → None

Append data to the median computer.

Parameters:

• data – The data to append to the median computer. Must have shape full_shape[1:].

• idx – The index at which to append the data. Must be between 0 and full_shape[0]. Required if using in-memory median computation.

evaluate() → ndarray

Compute the median data from the input data.

Returns:

The median data computed from the input data.

Return type:

np.ndarray