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.