Is your feature request related to a problem? Please describe.

Trajectory analyses are often affected by missing data (NaNs), which can bias metrics or invalidate results. Manually filtering for regions with fewer NaNs is tedious and subjective. An automated filter would improve reproducibility and streamline analysis.

Describe the solution you'd like

A filter function, e.g., filter_valid_regions, that automatically selects and returns the longest (or best) contiguous region(s) of a trajectory (or batch of trajectories) with minimal or no NaNs, based on a user-defined threshold for missing data. This filter should work at the dataset level and return a filtered DataArray or Dataset ready for downstream analysis.

Example usage:

import xarray as xr
from movement.filters import filter_valid_regions

# Assume 'position' is an xarray.DataArray with shape (time, space, keypoint, individual)
# Filter to region(s) with <20% NaNs
```python
filtered = filter_valid_regions(
    position,
    max_nan_fraction=0.2,
    min_length=100,
    dim="time",
)

# Now you can compute metrics like straightness_index on the filtered data
from movement.kinematics import straightness_index
si = straightness_index(position_filtered)

Describe alternatives you've considered

Manual masking or slicing, which is error-prone and not scalable for large datasets.

Additional context

This filter would complement the new straightness_index and other trajectory metrics, enabling robust, automated motif discovery and statistical analysis even in the presence of missing data.

References

• trajr R package — Implements trajectory metrics and region-based filtering utilities
  CRAN homepage: https://cran.r-project.org/package=trajr
  Documentation PDF: https://cran.r-project.org/web/packages/trajr/trajr.pdf

• CRCNS OC-1 dataset
  Open-field mouse trajectory dataset with real-world missing data, widely used in systems neuroscience and behavioral analysis.
  https://crcns.org/data-sets/oc/oc-1

• Related Movement issues and PRs

  • Add function for computing straightness index #517 — Add function for computing straightness index
  • Add trajectory complexity measures - implement straightness index #514 — Initial PR exploring path complexity metrics
  • Methods to measure complexity of trajectories (paths) #406 — Methods to measure complexity of trajectories
  • feat: Add straightness index metrics and behavioral trajectory analysis example #834 — Adds a native straightness_index utility; highlights the need for robust preprocessing of NaN-heavy trajectories

• Educational tutorial: “Animal tracking with trajr”
  Blog-style walkthrough demonstrating trajectory metrics and region filtering
  (Michael Crump).