How vonguyenleduy/stat-dsm works
Stat-DSM is a command-line research program for finding statistically discriminative sub-trajectories in labeled hurricane trajectories. Its workflow uses PostgreSQL trajectory data, searches for spatial neighbors and candidate subsequences, evaluates candidates against permuted labels, and reports an adjusted significance level. Most behavior is concentrated in one Python file, so the graph separates its domain stages without implying they are independent services. The excerpts show database access and parts of the mining/statistical logic; some orchestration links are documented only at a high level.
Main entry points: stat_dsm.py Command-line runner · max_euclidean.py Top-k distance
Components 9
Sub-trajectory analysis
- Command-line runner
stat_dsm.py - Trajectory data access
stat_dsm.py - Candidate mining
stat_dsm.py - Neighbor expansion
stat_dsm.py - Permutation testing
stat_dsm.py - Adjusted significance
stat_dsm.py
Distance calculation
- Top-k distance
max_euclidean.py
Other components
- Researcher
- PostgreSQL database
Connections 10
- Researcher → Command-line runner: runs program
README.md - Command-line runner → Trajectory data access: loads data
stat_dsm.py - Trajectory data access → PostgreSQL database: queries trajectories
stat_dsm.py - Candidate mining → Trajectory data access: reads trajectories
stat_dsm.py - Candidate mining → Neighbor expansion: expands candidates
stat_dsm.py - Neighbor expansion → Top-k distance: measures proximity
stat_dsm.py - Candidate mining → Permutation testing: tests candidates
stat_dsm.py - Command-line runner → Permutation testing: prepares label permutations
stat_dsm.py - Permutation testing → Adjusted significance: updates threshold
stat_dsm.py - Command-line runner → Adjusted significance: reports result
README.md