v25.0.0 - Major Feature Update: Deep Learning Integration, 3D Support, and Performance Improvements

🚀 Major Features

Deep Learning Integration

• Data Augmentation: 5 methods (rotate, noise, reverse, scale, subsample) for training robust models
• Sequence Processing: Padding, truncation, and normalization for batch training
• PyTorch Integration: Tensor conversion with graceful numpy fallback
• Feature Extraction: ML-ready features (displacement, speed, turn_angle, heading, acceleration)
• Dataset Utilities: train_test_split with shuffle and random_state support
• Demo Notebook: Complete example using public jaguar tracking dataset

3D Trajectory Support

• Full x, y, z coordinate handling throughout the library
• 3D distance, displacement, and plotting capabilities
• Backward compatible - 2D trajectories work exactly as before

GPS/Lat-Long Support

• Convert GPS coordinates to local x, y in meters (Haversine formula)
• Create trajectories directly from lat/lon arrays
• Preserves original GPS coordinates alongside local coordinates

Visualization Enhancements

• Interactive 2D/3D plots with plotly (zoom, pan, rotate)
• Heatmap visualization showing time spent in each location
• Speed and acceleration profile plots
• Comprehensive 4-panel trajectory analysis

Performance Optimizations

• Parallel batch processing using joblib (10-100x speedup for large datasets)
• Replaced inefficient np.append() calls (major speedup in calc_flow_angles)
• Caching mechanism for expensive computations

🔧 Bug Fixes & Improvements

• Fixed pandas 2.0+ compatibility (deprecated pd.datetime)
• Fixed lambda closure bug in date parsing
• Fixed missing inplace=True operations (6 critical silent failures)
• Replaced 18 generic exceptions with specific types
• Made PyTorch, plotly, and joblib optional dependencies with graceful fallbacks
• Fixed scipy deprecated window name (hanning → hann)

⚠️ Breaking Changes

• Python 3.8+ Required: Dropped Python 3.6 and 3.7 support (EOL versions)
• Updated minimum dependencies for better compatibility

📦 Installation

```bash

Core package

pip install traja

With deep learning extras (PyTorch, plotly, joblib)

pip install traja[dl]

All optional dependencies

pip install traja[all]
```

📚 Documentation

• Documentation: https://traja.readthedocs.io/en/latest/
• Deep Learning Guide: https://traja.readthedocs.io/en/latest/deep_learning.html
• API Reference: https://traja.readthedocs.io/en/latest/reference.html

🧪 Testing

• 46 new tests added for all features
• All tests pass on Python 3.8-3.12
• Tests work with and without optional dependencies

See CHANGELOG.md for complete details.

v22.0.0

Fix trip grid rotation error

v0.2.13

Paper update

v0.2.12

Paper update

v0.2.11

Paper update

v0.2.10

Paper update

v.0.2.9

Bump version

v0.2.8

Fix subpackages import

v0.2.7

Latest version

v0.1.1

Add v0.1.1 tag for PyPI