evan/transit-sim
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

population and job data for map, add visualization of data

Browse Source
This commit is contained in:
Evan Scamehorn
2025-12-02 14:06:24 -06:00
parent 852016088b
commit a3e057ef37
6 changed files with 238 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

145
data-pipeline/generate_city.py
View File

@@ -4,7 +4,9 @@ import os
import networkx as nx
from shapely.ops import unary_union
from shapely.geometry import Polygon, MultiPolygon
from scipy.spatial import cKDTree
import re
import numpy as np
# ==========================================
# 1. Configuration
@@ -29,6 +31,12 @@ DEFAULT_WIDTHS = {
}
# Census/Zoning Simulation Settings
# Approximate density per cubic meter of building volume
POP_DENSITY_FACTOR = 0.05 # People per m3 (Residential)
JOB_DENSITY_FACTOR = 0.08 # Jobs per m3 (Commercial)
# ==========================================
# 2. Helpers
# ==========================================
@@ -56,7 +64,6 @@ def get_height(row):
def estimate_road_width(row):
"""Estimates width with US-unit safety checks."""
# 1. Explicit width tag
for key in ["width", "width:carriageway", "est_width"]:
if key in row and str(row[key]) != "nan":
val_str = str(row[key]).lower()
@@ -66,13 +73,12 @@ def estimate_road_width(row):
val = float(nums[0])
if "'" in val_str or "ft" in val_str or "feet" in val_str:
val *= 0.3048
elif val > 50: # Sanity check for feet without units
elif val > 50:
val *= 0.3048
return val
except:
pass
# 2. Lanes
if "lanes" in row and str(row["lanes"]) != "nan":
try:
clean = re.findall(r"\d+", str(row["lanes"]))
@@ -83,18 +89,81 @@ def estimate_road_width(row):
except:
pass
# 3. Default based on type
highway = row.get("highway", "residential")
if isinstance(highway, list):
highway = highway[0]
return DEFAULT_WIDTHS.get(highway, 4.0)
def classify_building(row, height, area):
"""
Classifies a building as Residential (Pop) or Commercial (Jobs)
and estimates the count based on volume.
"""
b_type = str(row.get("building", "yes")).lower()
amenity = str(row.get("amenity", "")).lower()
office = str(row.get("office", "")).lower()
shop = str(row.get("shop", "")).lower()
volume = area * height
# Lists of tags
residential_tags = [
"apartments",
"residential",
"house",
"detached",
"terrace",
"dormitory",
"hotel",
]
commercial_tags = [
"commercial",
"office",
"retail",
"industrial",
"university",
"school",
"hospital",
"public",
]
is_res = any(t in b_type for t in residential_tags)
is_com = (
any(t in b_type for t in commercial_tags)
or (amenity != "nan" and amenity != "")
or (office != "nan" and office != "")
or (shop != "nan" and shop != "")
)
# Default logic if generic "yes"
if not is_res and not is_com:
# Small buildings likely houses, big generic likely commercial in city center
if volume > 5000:
is_com = True
else:
is_res = True
pop = 0
jobs = 0
category = "none"
density_score = 0
if is_res:
pop = round(volume * POP_DENSITY_FACTOR)
category = "residential"
density_score = min(1.0, pop / 500) # Normalize for color (0-1)
elif is_com:
jobs = round(volume * JOB_DENSITY_FACTOR)
category = "commercial"
density_score = min(1.0, jobs / 1000) # Normalize for color (0-1)
return category, density_score, pop, jobs
def parse_geometry(geom, center_x, center_y):
"""Parses geometry into {outer, holes} structure."""
if geom.is_empty:
return []
polys = []
if geom.geom_type == "Polygon":
source_geoms = [geom]
@@ -116,12 +185,10 @@ def parse_geometry(geom, center_x, center_y):
]
holes.append(hole_coords)
polys.append({"outer": outer, "holes": holes})
return polys
def parse_line_points(geom, center_x, center_y):
"""Simple parser for LineStrings (Routing Graph)."""
if geom.geom_type == "LineString":
return [
[round(x - center_x, 2), round(y - center_y, 2)] for x, y in geom.coords
@@ -139,6 +206,9 @@ tags_visual = {
"natural": ["water", "bay"],
"leisure": ["park", "garden"],
"landuse": ["grass", "forest", "park"],
"amenity": True, # Fetch amenities to help classify jobs
"office": True,
"shop": True,
}
gdf_visual = ox.features.features_from_address(PLACE_NAME, tags=tags_visual, dist=DIST)
@@ -158,24 +228,42 @@ center_y = gdf_visual.geometry.centroid.y.mean()
output_visual = {"buildings": [], "water": [], "parks": [], "roads": []}
output_routing = {"nodes": {}, "edges": []}
print("3. Processing Visual Layers...")
# We will store building data to map it to graph nodes later
building_data_points = [] # (x, y, pop, jobs)
print("3. Processing Visual Layers & Census Simulation...")
for idx, row in gdf_visual.iterrows():
polygons = parse_geometry(row.geometry, center_x, center_y)
for poly_data in polygons:
# 1. Buildings
# 1. Buildings (With Zoning Logic)
if "building" in row and str(row["building"]) != "nan":
height = get_height(row)
area = row.geometry.area
# Zoning / Census Simulation
cat, score, pop, jobs = classify_building(row, height, area)
# Store centroid for graph mapping
cx = row.geometry.centroid.x - center_x
cy = row.geometry.centroid.y - center_y
building_data_points.append([cx, cy, pop, jobs])
output_visual["buildings"].append(
{"shape": poly_data, "height": get_height(row)}
{
"shape": poly_data,
"height": height,
"data": {"type": cat, "density": score, "pop": pop, "jobs": jobs},
}
)
# 2. Water (Explicit check for NaN)
# 2. Water
elif ("natural" in row and str(row["natural"]) != "nan") or (
"water" in row and str(row["water"]) != "nan"
):
output_visual["water"].append({"shape": poly_data})
# 3. Parks (Fallback)
# 3. Parks
else:
output_visual["parks"].append({"shape": poly_data})
@@ -187,17 +275,40 @@ for idx, row in gdf_edges.iterrows():
road_polys.append(buffered)
if road_polys:
print(" Merging road polygons...")
merged_roads = unary_union(road_polys)
road_shapes = parse_geometry(merged_roads, center_x, center_y)
for shape in road_shapes:
output_visual["roads"].append({"shape": shape})
print("4. Processing Routing Graph...")
print("4. Mapping Census Data to Graph Nodes...")
# Create a KDTree of building centroids
if building_data_points:
b_coords = np.array([[b[0], b[1]] for b in building_data_points])
b_data = np.array([[b[2], b[3]] for b in building_data_points]) # pop, jobs
tree = cKDTree(b_coords)
for node_id, row in gdf_nodes.iterrows():
nx = row.geometry.x - center_x
ny = row.geometry.y - center_y
# Find all buildings within 100m of this node
if building_data_points:
indices = tree.query_ball_point([nx, ny], r=100)
if indices:
# Sum pop/jobs of nearby buildings
nearby_stats = np.sum(b_data[indices], axis=0)
node_pop = int(nearby_stats[0])
node_jobs = int(nearby_stats[1])
else:
node_pop, node_jobs = 0, 0
else:
node_pop, node_jobs = 0, 0
output_routing["nodes"][int(node_id)] = {
"x": round(row.geometry.x - center_x, 2),
"y": round(row.geometry.y - center_y, 2),
"x": round(nx, 2),
"y": round(ny, 2),
"pop": node_pop, # Store for gameplay later
"jobs": node_jobs,
}
for u, v, k in G.edges(keys=True):