Skip to content

Handle time gaps in ML classifier models (e.g., ActiGraph Idle Sleep Mode) #76

Description

@jhmigueles

When processing .gt3x files with ActiGraph Idle Sleep Mode, time gaps are present in the data which are handled with last observation carrief forwared after normalizing the last epoch to 1 g (i.e., assuming no movement). This creates synthetic flatline epochs (0 magnitude, 0 variability). Because ML models were never trained on zero-variance inputs, some of them misclassify these idle epochs as active classes (e.g., lightGames) rather than sedentary.

Proposal: Rule-based Override to flag those affected epochs and force them to sedentary in classify() if they are not flagged as non-wear or sleep.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions