Skip to content

Developer notes

Development environment

If you want to create your own metrics or filters, we recommend setting up a local development environment.

The recommended way to set up your local environment to run Pydre is to use Rye:

  1. Install Rye.
  2. Either clone the Git repository to a local directory or unzip a release package of Pydre
  3. Navigate to the Pydre directory in the terminal.
  4. Run rye sync

This will download the appropriate python packages needed to run Pydre.

DriveData objects

This is the primary unit of data storage for the module. DriveData objects are initially created from dat files, and then they are split by the ROI functions.

  • SubjectID: Unique identifier for this object. Any file loaded into a DriveData object should ONLY be data from this subject number, however, this is not currently enforced
  • roi: Singular string denoting the region of interest of the particular DriveData. There can currently only be one region of interest per DriveData object
  • data: Polars dataframe containing the rows of the drive data
  • sourcefilename: The filename of where the data originally came from

Project objects

This is where the processing actually takes place. A project is constructed by reading in a project file (either JSON or TOML).

Tips for Writing New Metrics

Metrics are used to calculate values from the data, and the calculations are done after the data is filtered.

Everytime you want to make a new metric, you need to use the @registerMetric decorator followed by the function definition.

Generally, every metric begins with the same two lines:

required_col = [array of column names]  
try:
    drivedata.checkColumnsNumeric(required_col)
except pl.exceptions.PolarsError:
    return None
This is to ensure that the columns you need in order to calculate your metric are part of the dataset.

As metrics calculate some value, every metric has a return statement.
If no output is specified, the return value gets outputted in an out.csv file under a column titled the metrics name.
If the data is sectioned into multiple regions of interests (rois), the metric will be processed on each roi and produce a row for each roi in the output file.