Examples

Basic Trajectory Processing

import pandas as pd
import pyehicle as pye

# Load and compress trajectory
df = pd.read_csv('trajectory.csv')
compressed = pye.preprocessing.spatio_temporal_compress(df, spatial_radius_km=0.01, time_threshold_s=30)
print(f"Reduced from {len(df)} to {len(compressed)} points")

# Map-match to road network
matched = pye.preprocessing.leuven(compressed, lat_col='lat', lon_col='lon', max_dist=50)
print(f"Matched {len(matched)} points to road network")

Complete Reconstruction Pipeline

import pandas as pd
import pyehicle as pye

# Load data
df = pd.read_csv('raw_trajectory.csv')

# Preprocessing
compressed = pye.preprocessing.spatio_temporal_compress(df, spatial_radius_km=0.01, time_threshold_s=30)
matched = pye.preprocessing.leuven(compressed, lat_col='lat', lon_col='lon', max_dist=50)
filtered = pye.preprocessing.kalman_aeqd_filter(matched, lat_col='lat', lon_col='lon', time_col='time')

# Load road network
road_network, geometries, spatial_index = pye.utilities.load_road_network(
    pbf_file_path='map.osm.pbf',
    bbox=(24.0, 56.9, 24.2, 57.0)
)

# Reconstruction
refined = pye.reconstructing.refine_trajectory(filtered, road_network, max_node_distance=10, time_col='time')
final_trajectory = pye.reconstructing.curve_interpolation(refined, road_network, lower_threshold=20, upper_threshold=80, time_col='time')

# Evaluate
ground_truth = pd.read_csv('ground_truth.csv')
f1_score = pye.utilities.f1(final_trajectory, ground_truth)
precision = pye.utilities.precision(final_trajectory, ground_truth)
recall = pye.utilities.recall(final_trajectory, ground_truth)

print(f"F1: {f1_score:.3f} | Precision: {precision:.3f} | Recall: {recall:.3f}")

# Visualize
pye.utilities.visualization.multiple(
    df_list=[matched, filtered, final_trajectory],
    names=['Map-Matched', 'Kalman Filtered', 'Reconstructed'],
    show_in_browser=True,
    cmap='tab10'
)

Combining Topologically Equivalent Trajectories

import pandas as pd
import pyehicle as pye

# Load multiple recordings of the same bus route
route_66_run1 = pd.read_csv('bus_route_66_monday.csv')
route_66_run2 = pd.read_csv('bus_route_66_tuesday.csv')
route_66_run3 = pd.read_csv('bus_route_66_wednesday.csv')

# Preprocess each recording independently
processed_runs = []
for i, run in enumerate([route_66_run1, route_66_run2, route_66_run3]):
    compressed = pye.preprocessing.spatio_temporal_compress(run, spatial_radius_km=0.01, time_threshold_s=30)
    matched = pye.preprocessing.leuven(compressed, lat_col='lat', lon_col='lon', max_dist=50)
    filtered = pye.preprocessing.kalman_aeqd_filter(matched, lat_col='lat', lon_col='lon', time_col='time')
    processed_runs.append(filtered)
    print(f"Run {i+1}: {len(run)} → {len(filtered)} points")

# Combine all runs into a single canonical route
canonical_route = pye.reconstructing.trajectory_combiner(
    processed_runs,
    lat_col='lat',
    lon_col='lon',
    time_col='time'
)
print(f"Combined into canonical route with {len(canonical_route)} points")

# Load road network and refine the canonical route
road_network, geometries, spatial_index = pye.utilities.load_road_network(
    pbf_file_path='map.osm.pbf',
    bbox=(24.0, 56.9, 24.2, 57.0)
)

refined = pye.reconstructing.refine_trajectory(canonical_route, road_network, max_node_distance=10, time_col='time')
final_route = pye.reconstructing.curve_interpolation(refined, road_network, lower_threshold=20, upper_threshold=80, max_node_distance=10, time_col='time')

# Visualize the final canonical route
pye.utilities.visualization.single(final_route, name='Bus Route 66 - Canonical Trajectory', show_in_browser=True)