spatiotemporal-data

tsl: a PyTorch library for processing spatiotemporal data.

PAMI is a Python library containing 100+ algorithms to discover useful patterns in various databases across multiple computing platforms. (Active)

[KDD 2024] "ImputeFormer: Low Rankness-Induced Transformers for Generalizable Spatiotemporal Imputation"

Given a shapefile with time-annotated vector objects (e.g., building footprints + construction year), this script will automatically create an animated GIF illustrating the dynamics for a user-specified period of time

Repository for ''Contextualizing MLP-Mixers Spatiotemporally for Urban Data Forecast at Scale''

Laplacian-enhanced tensor learning for large-scale spatiotemporal traffic data kriging (estimation)

Conditional Latent Autoregressive Recurrent Model (CLARM) for learning spatiotemporal dynamics

Bayesian spatial regression with Meshed Gaussian Process

A unified Python workflow for POD, MPOD, DMD, SPOD, PSD-POD, BSMD, and ST-POD on spatiotemporal data.

Official repository for the paper "Back to the Future: GNN-based NO2 Forecasting via Future Covariates" , IGARSS 2024

"Evaluation procedures for forecasting with spatio-temporal data" -- Mathematics (MDPI) 2021

Applied data science research adapting UK Met Office rain-cell tracking to Taiwan radar and 2.8M rain-gauge observations

Interactive benchmark for comparing trajectory embeddings through intrinsic and extrinsic evaluation for location prediction.

Reproducible pipeline for cleaning Kansas crash records (KCARS/KORA, 2013-2024) into ML- and geospatial-analysis-ready data for traffic safety research. Includes sample dataset, full documentation, and an interactive filter tool.

This is the GitHub repository for the preprint https://arxiv.org/pdf/2502.07111.

Pattern specifications: evolving phase-aware observers for agnostic structure discovery in spatiotemporal fields (JSA submission) — code, experiments, results

Time-resolved, multi-source observational field generator. Ingests heterogeneous real-world data streams, validates against source-faithful schemas, and preserves atomic, geo-resolved observations with full temporal fidelity for downstream collation, analysis, and latent variable construction.

Reproducible soccer tracking-data research on off-ball movement and localized defensive reorganization across Metrica, IDSSE/DFL, and SkillCorner data.

🖼️ SpatioTemporal Visual Storyteller (STVS) 🖼️