Introduction

spudtr provides some special purpose Python 3 functions for transforming pandas.DataFrame objects organized in a particular way. It leans heavily on numpy and scipy.

The emphasis is operations on multichannel synchronous discrete time series instrument recordings, such as multi-sensor arrays.

The origin is experimental EEG recordings where the interest is in the brain activity before, during, and after stimulus and response events which are logged and time-stamped on recording channels concurrently with the EEG. Additional data that plays a role in analysis and modeling such as quality control codes and other experimental variables is merged in as additional columns alongside the other data streams which makes a tidy structure for feeding in the analysis pipelines of primary interest.

This leads naturally to snipping apart longish data logger recordings into a 2-D tabular data structure that contains a vertical stack of fixed-length intervals or “epochs” of recordings, indexed by epoch and time stamped (rows) \(\times\) parallel data streams (columns).

This is the spudtr epochs data format (pandas.DataFrame) that most of the functions ingest and transform. Some helper functions like FIR filtering are agnostic about epochs and just operate on time series data.

The User Guide has worked examples for some common use cases.

The Reference lists all the available functions and documentation, with links to the source with more helper functions under the hood.