Traja

Traja is a Python library for trajectory analysis. It extends the capability of pandas DataFrame specific for animal trajectory analysis in 2D, and provides convenient interfaces to other geometric analysis packages (eg, R and shapely).

Introduction

The traja Python package is a toolkit for the numerical characterization and analysis of the trajectories of moving animals. Trajectory analysis is applicable in fields as diverse as optimal foraging theory, migration, and behavioral mimicry (e.g. for verifying similarities in locomotion). A trajectory is simply a record of the path followed by a moving animal. Traja operates on trajectories in the form of a series of locations (as x, y coordinates) with times. Trajectories may be obtained by any method which provides this information, including manual tracking, radio telemetry, GPS tracking, and motion tracking from videos.

The goal of this package (and this document) is to aid biological researchers, who may not have extensive experience with Python, to analyze trajectories without being restricted by a limited knowledge of Python or programming. However, a basic understanding of Python is useful.

If you use traja in your publications, please cite the repo

@software{justin_shenk_2019_3237827,
  author       = {Justin Shenk and
                  the Traja development team},
  title        = {justinshenk/traja},
  month        = jun,
  year         = 2019,
  publisher    = {Zenodo},
  version      = {latest},
  doi          = {10.5281/zenodo.3237827},
  url          = {https://doi.org/10.5281/zenodo.3237827}
}

Installation and setup

To install traja with conda, run

conda install -c conda-forge traja

or with pip

pip install traja.

Import traja into your Python script or via the Python command-line with import traja.

Trajectories with traja

Traja stores trajectories in pandas DataFrames, allowing any pandas functions to be used.

Load trajectory with x, y and time coordinates:

import traja

df = traja.read_file('coords.csv')

Once a DataFrame is loaded, use the .traja accessor to access the visualization and analysis methods:

df.traja.plot(title='Cage trajectory')

Analyze Trajectory

The following functions are available via traja.trajectory.[method]

Function	Description
calc_derivatives	Calculate derivatives of x, y values
calc_turn_angles	Calculate turn angles with regard to x-axis
transitions	Calculate first-order Markov model for transitions between grid bins
generate	Generate random walk
resample_time	Resample to consistent step_time intervals
rediscretize_points	Rediscretize points to given step length

For up-to-date documentation, see https://traja.readthedocs.io.

Deep Learning Integration

Traja provides production-ready features for training neural networks on trajectory data:

Data Augmentation - Create training variations for robust models:

# Rotation, noise, scaling, reversal, subsampling
rotated = df.traja.augment_rotate(angle=45)
noisy = df.traja.augment_noise(sigma=0.1)
scaled = df.traja.augment_scale(factor=1.5)

Sequence Processing - Standardize trajectory lengths for batching:

# Pad or truncate to fixed length
padded = df.traja.pad_trajectory(target_length=200, mode='edge')
truncated = df.traja.truncate_trajectory(target_length=100, mode='random')
normalized = df.traja.normalize_trajectory()

Feature Extraction - Generate ML-ready features:

# Extract displacement, speed, turn_angle, heading, acceleration
features = df.traja.extract_features()

PyTorch Integration - Seamless tensor conversion:

tensor = df.traja.to_tensor()  # Convert to PyTorch tensor

Dataset Utilities - Train/val/test splitting:

trajectories = [traja.generate(n=100) for _ in range(50)]
train, val, test = traja.trajectory.train_test_split(
    trajectories, train_size=0.7, val_size=0.15, test_size=0.15
)

3D Support - All features work with x, y, z coordinates:

df_3d = traja.TrajaDataFrame({'x': x, 'y': y, 'z': z})
tensor_3d = df_3d.traja.to_tensor()  # Shape: (n_points, 3)

GPS/Lat-Long Support - Work with GPS coordinates:

traj = traja.from_latlon(lat, lon)  # Convert GPS to local x,y

Visualization Enhancements - Better trajectory analysis and exploration:

# Interactive plots with plotly
fig = df.traja.plot_interactive()  # Zoom, pan, rotate

# Heatmap showing time spent in locations
df.traja.plot_heatmap(bins=50)

# Speed and acceleration profiles
df.traja.plot_speed()
df.traja.plot_acceleration()

# Comprehensive 4-panel analysis
df.traja.plot_trajectory_components()

Performance Optimization - Fast parallel processing:

# Process 100 trajectories in parallel
trajectories = [traja.generate(n=1000) for _ in range(100)]
results = traja.trajectory.batch_process(
    trajectories,
    lambda t: t.traja.normalize_trajectory(),
    n_jobs=-1  # Use all CPUs
)

See the Deep Learning documentation and examples/deep_learning_demo.ipynb for complete examples.

Random walk

Generate random walks with

df = traja.generate(n=1000, step_length=2)
df.traja.plot()

Resample time

traja.trajectory.resample_time allows resampling trajectories by a step_time.

Flow Plotting

df = traja.generate()
traja.plot_surface(df)

traja.plot_quiver(df, bins=32)

traja.plot_contour(df, filled=False, quiver=False, bins=32)

traja.plot_contour(df, filled=False, quiver=False, bins=32)

traja.plot_contour(df, bins=32, contourfplot_kws={'cmap':'coolwarm'})

Acknowledgements

traja code implementation and analytical methods (particularly rediscretize_points) are heavily inspired by Jim McLean's R package trajr. Many thanks to Jim for his feedback.