Untitled project

!wget https://analytics.urbanpulse.de/files/hackathon2019/Darmstadt_20181229-20190111.csv

import pandas as pd import geopandas as gpd from shapely.geometry import Point # Read the data into a Pandas DataFrame df = pd.read_csv('Darmstadt_20181229-20190111.csv') #Create Point objects for each lat, lon pair geometry = [Point(xy) for xy in zip(df.lon, df.lat)] #Create a GeoDataFrame with Point objects and other columns crs = 'EPSG:3857' geo_data = gpd.GeoDataFrame(df, crs=crs, geometry=geometry) #Group the GeoDataFrame by small geo-referenced bins with a side length of 20-50 meters bin_size = 0.02 # change to 0.05 if you prefer a larger bin size geo_data['x'] = ((geo_data.geometry.x - geo_data.geometry.x.min()) / bin_size) geo_data['y'] = ((geo_data.geometry.y - geo_data.geometry.y.min()) / bin_size) #Calculate the 85th percentile of speed for each bin grouped = geo_data.groupby(['x', 'y'])['speed'].quantile(0.85) bin_centers = [] percentile_speeds = [] for (x, y), speed in grouped.items(): bin_centers.append(Point((x+0.5)*bin_size+geo_data.geometry.x.min(), (y+0.5)*bin_size+geo_data.geometry.y.min())) percentile_speeds.append(speed) #Create a new GeoDataFrame with only the bin_id and percentile_85 columns result = gpd.GeoDataFrame({'percentile_speed': percentile_speeds}, geometry=bin_centers, crs=crs) #Extract lat and lon from the centerpoint result["long"] = result['geometry'].map(lambda p: p.x) result["lat"] = result['geometry'].map(lambda p: p.y) result.to_file('processed_result.json')