Build, automate, and scale spatiotemporal movement pipelines.

A field manual for mobility data scientists, urban analysts, Python GIS developers, and logistics engineering teams. Every guide is focused on coordinate and temporal precision, production-ready Python, and the messy reality of debugging real-world movement data — not theoretical GIS overviews.

Trajectory segmentation, stay-point detection, time-window mapping, GPS error correction, and pipeline synchronization — written so you can drop patterns straight into your stack and ship reliable analytics from day one.

Spatiotemporal Data Foundations & Structures

Data models, coordinate systems, sampling architecture, GPS error handling and storage patterns for movement data.

Open the guide

Movement Pattern Extraction & Trajectory Analysis

Segmentation, stay-point detection, kinematic profiling, directionality analysis and change detection at scale.

Open the guide

Temporal Aggregation & Window Mapping

Turn asynchronous telemetry into structured spatiotemporal matrices: time binning, rolling stats, gap filling, seasonal alignment.

Open the guide

Spatial Indexing & Density Aggregation

Discrete global grids, map-matching, origin-destination matrices and kernel density surfaces — turn cleaned trajectories into aggregated spatial signal.

Open the guide
Start here

Frequently reached guides

These pages are the most direct entry points for common real-world problems. Each is a self-contained, executable guide you can work through in an afternoon.

Handling GPS Drift in Raw Trajectory Logs Temporal regularization, kinematic thresholding, and Savitzky-Golay smoothing. DBSCAN for Stay-Point Clustering in Python scikit-learn DBSCAN with Haversine metric, radian conversion, and temporal filtering. CRS Transformations in Movement Data Safe pyproj pipelines, 3D coordinate handling, and output validation. Kalman Filter Gap Filling for GPS Dropouts 2D constant-velocity Kalman filter for irregular timestamps with production tuning. Classifying Walk, Bike and Car from Speed Features A runnable baseline classifier, and why the train-test split has to be by entity. Structuring Trajectory Data in GeoPandas Long-format GeoDataFrame schema, CRS setup, spatial indexing, and segmentation. H3 vs S2 vs Quadkey for Density Grids Choosing a discrete global grid for mobility density: distortion, joins, and cost. Feeding Cleaned GPS into an HMM Map-Matcher Wiring drift-corrected trajectories into a Hidden Markov Model map-matching stage.
New

Recently added

Five new subject areas, each with its own guides: inferring how somebody travelled, telling a sensor artefact from a real anomaly, windowing a live feed, making space-and-time queries fast, and releasing movement data without releasing the people in it.

Transport Mode Inference Walk, cycle, road or rail from the shape of a segment — and where speed alone stops working. Movement Anomaly Detection Physics-based rejection, route-deviation scoring, and an alert queue somebody can triage. Streaming Window Processing Event time, watermarks sized from measured delivery lag, and bounded window state. Spatiotemporal Query Indexing Why a spatial index alone is slow on trajectories, and the pruning funnel that fixes it. Movement Data Privacy & Anonymization Endpoint truncation, generalisation, k-anonymity — and measuring what each one costs. Partitioning Trajectory Parquet by Time and Space Partition on time in the directory tree, on space in the sort order — and why. Detecting Harsh Braking and Acceleration Events Speed-dependent thresholds, a duration requirement, and the exposure denominator. Using HDOP and Satellite Count to Weight GPS Fixes Weight the fixes instead of gating them, and feed the result to the filter.
What's inside

Four sections, twenty-five topics, engineered end-to-end

Each section is broken into narrowly scoped topics, and every topic carries the in-depth guides that make it usable. All eighty-five pages contain executable Python you can lift into your own stack, with calibration tables, validation blocks that fail loudly, and links to the foundational concepts underneath.

Spatiotemporal Data Foundations & Structures

Data models, coordinate systems, sampling architecture, GPS error handling and storage patterns for movement data.

Coordinate Reference System Mapping Read the guide → GPS Precision & Error Handling Read the guide → Movement Data Privacy & Anonymization Read the guide → Sampling Rate Optimization for Movement Trajectories Read the guide → Spatial Storage Formats Read the guide → Time-Series Synchronization Strategies for Movement Data Read the guide → Trajectory Object Design Patterns Read the guide →

Movement Pattern Extraction & Trajectory Analysis

Segmentation, stay-point detection, kinematic profiling, directionality analysis and change detection at scale.

Directionality & Turn Analysis: Quantifying Heading Changes in Movement Trajectories Read the guide → Movement Anomaly Detection Read the guide → Movement Similarity & Clustering Read the guide → Speed & Acceleration Profiling for Trajectory Data Read the guide → Stay-Point Detection Algorithms for Mobility Data Pipelines Read the guide → Trajectory Segmentation Read the guide → Transport Mode Inference Read the guide →

Temporal Aggregation & Window Mapping

Turn asynchronous telemetry into structured spatiotemporal matrices: time binning, rolling stats, gap filling, seasonal alignment.

Dynamic Time-Binning Strategies for Movement Telemetry Read the guide → Gap Filling in Sparse Trajectories Read the guide → Rolling Statistics for Mobility Metrics Read the guide → Seasonal & Cyclical Alignment for Movement Data Read the guide → Streaming Window Processing Read the guide → Time-Series Synchronization Strategies for Temporal Aggregation Read the guide →
Why this site

For the engineers shipping movement analytics

Coordinate & temporal precision UTC-first ingestion, projected CRS for metric work, deterministic window boundaries.
Production-ready Python Polars, GeoPandas, Shapely, pyproj, MovingPandas — patterns that scale past notebooks.
Real debugging GPS drift, multipath, signal loss, clock jitter, projection mismatches — diagnosed and fixed.
Pipeline-shaped Every guide ends with validation gates, observability, and operational SLAs.