This library and script collection encompasses a wide variety of tooling that I developed during my dissertation on mobile positioning data processing for urban system modeling.

Some of the scripts belong to workflows presented in one of my articles; those are grouped under their thematic descriptions. Other scripts and utilities are described further below.

Šimbera, J., Aasa, A.: Commuting phones: Moving on from enumeration to mobile positioning in commuting statistics. Journal of Urban Technology (in review).

This article shows how to convert mobile positioning-derived anchor points of individual people to aggregate spatial interactions (commuting flows), which are of great importance in spatial analysis.

The main body of the code is located in the mobilib.relations module. Run the scripts as follows:

• anchors_to_rels to get interactions from anchor points
  • calculate_eta might be used to calculate the self-interaction parameter eta, but it does not really help with anything
• transfer_rels to areally interpolate interactions between two sets of spatial units (such as from mobile network cells to administrative units)
  • transfer_table to create the required transfer table that specifies the fractions of source spatial units (mobile network cells) to be transferred to respective target spatial units (administrative units) using a weighting layer that specifies how the users are distributed within the source spatial units
  • transfer_table_aw to substitute transfer_table by a simpler areal weighting process that is less accurate but does not require the weighting layer
• (not covered in the article) functionality in mobilib.diffpriv to apply differential privacy anonymization to the obtained interaction values
• calibrate_rels to calibrate interaction values to census or other calibration data
• rels_to_lines to visualize the created interactions into a line geometry spatial file
  • representative_points converts a spatial file to a point file of representative points as provided by the Shapely algorithm (useful if you only have a polygon file to represent interacting units)
  • weighted_centroids can be used to do the same but in a more advanced fashion, using a weighting layer (such as the layer of built-up areas)

Šimbera, J., Aasa, A.: Hierarchical settlement system model: making polycentricity work with central place theory. Urban Studies (in review).

This article presents a way to model relationships and hierarchies in the settlement system using a model with three developmental stages. Here is how to create such a model from spatial interactions, evaluate and visualize it.

The main body of the code is located in the mobilib.hssm module. Run the scripts as follows:

• build_hssm to build the hierarchical settlement system model and store it in a table
• hssm_to_regions to transform the model into a set of functional regions with defined criteria (not currently operational)
  • show_zipf_gaps may reveal which population criteria might be the most meaningful for the given settlement system

Komárek, M., Marada, M., Šimbera, J. (2023): Metropolitní polynodální jádra jako základ nové regionální organizace. Geografie, 128, 1, 49−74. https://doi.org/10.37040/geografie.2023.004

The main body of the code is located in mobilib.region where regions can be formed by aggregation, and in mobilib.region_measure where arbitrary groupings of units can be measured based on functional region criteria. There are the following scripts:

• add_distances to make interactions distance-aware
• flow_intensity_indices to calculate some interaction metrics such as counts of commuters per kilometer of distance
• delimit_multicores to find groupings of units that form a core of a common region
• delimit_regions which annotates units by region ID they belong to, based on a sequential (CURDS-like) functional region delimitation procedure
  • show_zipf_gaps may also help here (see above)
• measure_regions to compute various metrics on such regions
• summarize_unit_rels to compute metrics on units based on their regional assignment
• dissolve_areas to turn these region annotations and unit geometries into region polygons with proper attributes
• eliminate_exclaves to eliminate exclaves of these region polygons by merging them with neighboring regions

• calibrate and calibrate_batch are counterparts of calibrate_rels for non-interaction data such as population counts.
• equalize_polygons makes a polygon set approximate a given uniform surface area by aggregation or splitting; this was meant as an extension of the HSSM study to compensate for MAUP
• eval_regression_error evaluates the regression error in given data files by producing HTML reports
• fix_polygon_neighbours fixes geometries in a polygon dataset so that their boundaries align nicely
• csv_to_shp and geo_to_csv convert between GDAL-compatible spatial files and custom semicolon-delimited CSVs with WKT or XY geometry
• lines_to_network prepares a road/rail/whatever line dataset to be routable using networkx
• raster_to_points converts a raster to a point layer
• signaling_trajectories smoothes out a point layer of trajectories for better inspection
• transfer_static is a variant of transfer_rels for non-interaction data (keyed by one identifier, e.g. mobile network cell populations)
• triangulate creates a neighbourhood interaction table of points by calculating which pairs of them are connected by an edge in the Delaunay triangulation
• voronoi_cells calculates ordinary Voronoi cells for a point dataset
• potential_raster, mobilib.potential and mobilib.raster contain old machinery to compute raster maps of geographic potential

• antenna_centroids and antenna_model for mobile network antenna coverage modeling (together with mobilib.antenna): experiment not finished, leaving here for future reference