Python starts when you choose Run.
The denominator inside a geospatial benchmark
This notebook reruns the attached offline analysis. Place all bundle files in one directory and run from that directory. The default path uses Python’s standard library only. Published measurements, metadata and illustrative numerical probes are identified in the article and provenance. No model or API calls occur.
All analysis assertions passed; figures and results regenerated.
{
"k_counts": {
"1": 10,
"2": 50,
"3": 15,
"4": 10
},
"template_counts": {
"k1_location_extreme_coordinate": 5,
"k1_datacommons_variable_ratio": 5,
"k3_places_dc_location_farthest_from_landmark": 5,
"k2_climate_risk_ee_admin_water_stress": 5,
"k2_climate_risk_buffer_water_stress": 4,
"k4_weather_location_places_dc_centroid_distance": 5,
"k2_flood_forecast_inundation_area_pct": 3,
"k2_climate_risk_buffer_temp_diff": 4,
"k2_climate_risk_ee_admin_days_above_temp": 5,
"k3_historical_weather_location_places_isolation": 10,
"k2_places_location_poi_pair_ratio": 5,
"k4_weather_location_places_dc_poi_extremes_distance": 5,
"k2_places_dc_ratio_and_poi_count": 5,
"k2_places_location_poi_density_comparison": 5,
"k2_climate_risk_city_buffer_water_stress_diff": 3,
"k2_places_dc_superlative_state_variant": 1,
"k2_places_dc_superlative_substate": 3,
"k2_places_dc_stat_to_poi_ratio": 4,
"k2_places_location_annulus_ring_pois": 3
},
"canonical_template_count": 18,
"canonical_template_counts": {
"k1_location_extreme_coordinate": 5,
"k1_datacommons_variable_ratio": 5,
"k3_places_dc_location_farthest_from_landmark": 5,
"k2_climate_risk_ee_admin_water_stress": 5,
"k2_climate_risk_buffer_water_stress": 4,
"k4_weather_location_places_dc_centroid_distance": 5,
"k2_flood_forecast_inundation_area_pct": 3,
"k2_climate_risk_buffer_temp_diff": 4,
"k2_climate_risk_ee_admin_days_above_temp": 5,
"k3_historical_weather_location_places_isolation": 10,
"k2_places_location_poi_pair_ratio": 5,
"k4_weather_location_places_dc_poi_extremes_distance": 5,
"k2_places_dc_ratio_and_poi_count": 5,
"k2_places_location_poi_density_comparison": 5,
"k2_climate_risk_city_buffer_water_stress_diff": 3,
"k2_places_dc_superlative_substate": 4,
"k2_places_dc_stat_to_poi_ratio": 4,
"k2_places_location_annulus_ring_pois": 3
},
"snapshot_dates": {
"2026-09-15": 73,
"2026-09-29": 1,
"2026-09-23": 11
},
"answer_slot_counts": {
"answer_slots": 179,
"numeric_slots": 113,
"string_slots": 66,
"zero_numeric_slots": 7,
"negative_numeric_slots": 5
},
"domain_solvability": {
"Natural Hazards & Crisis Resilience": {
"beyond_current_toolset": 7,
"solvable": 4
},
"Public Health": {
"beyond_current_toolset": 3,
"solvable": 2
},
"Site Selection & Urban Planning": {
"solvable": 4,
"beyond_current_toolset": 4
},
"Operational & Commercial Utilities": {
"beyond_current_toolset": 1
}
},
"solvability": {
"beyond_current_toolset": 15,
"solvable": 10
},
"workflow_list_items": {
"min": 13,
"median": 19,
"max": 40
},
"domain_weights_dev": {
"Natural Hazards & Crisis Resilience": 0.44,
"Public Health": 0.2,
"Site Selection & Urban Planning": 0.32,
"Operational & Commercial Utilities": 0.04
},
"domain_weights_test": {
"Natural Hazards & Crisis Resilience": 0.4605263157894737,
"Public Health": 0.2236842105263158,
"Site Selection & Urban Planning": 0.25,
"Operational & Commercial Utilities": 0.06578947368421052
},
"k_weights_dev": [
0.11764705882352941,
0.5882352941176471,
0.17647058823529413,
0.11764705882352941
],
"k_weights_test": [
0.14285714285714285,
0.47619047619047616,
0.23809523809523808,
0.14285714285714285
],
"k_composition_max_pp": 11.204481792717088,
"illustrative_profile": {
"rates": [
0.9,
0.8,
0.6,
0.4
],
"dev_percent": 72.94117647058825,
"heldout_mixture_percent": 70.95238095238095,
"gap_pp": 1.988795518207298
},
"numeric_probes": [
{
"case": "inside boundary",
"gold": 100,
"prediction": 110,
"score": 1.0
},
{
"case": "outside boundary",
"gold": 100,
"prediction": 110.01,
"score": 0.8998999999999999
},
{
"case": "zero exact",
"gold": 0,
"prediction": 0,
"score": 1.0
},
{
"case": "zero near miss",
"gold": 0,
"prediction": 0.01,
"score": 0.0
},
{
"case": "numeric JSON value",
"gold": 1000,
"prediction": 1000,
"score": 1.0
},
{
"case": "exponent in string",
"gold": 1000,
"prediction": "1e3",
"score": 0.0010000000000000009
},
{
"case": "comma in string",
"gold": 1000,
"prediction": "1,000",
"score": 0.0010000000000000009
},
{
"case": "negative boundary",
"gold": -100,
"prediction": -110,
"score": 1.0
}
],
"registry_status": "All 110 dev IDs unique; 85 parsed/raw answers agree; no agent scores computed."
}
Regenerated figures
The following rich SVG outputs come from files regenerated by the preceding code, not screenshots of an external result.
Python runs locally in this page. Saved results and figures are visible before execution.