Using TransBigData for Python Transportation Data Analysis and Visualization

1. Introduction

The TransBigData library is an open‑source Python package designed for efficient processing, analysis, and visualization of transportation spatio‑temporal big data such as taxi GPS, bike‑share, and bus GPS records. It provides concise, high‑performance functions that simplify complex data‑handling tasks.

2. Installation

Install the library via pip (or conda) with the following command:

<code>pip install -U transbigdata</code>

After installation, import the package in Python:

<code>import transbigdata as tbd</code>

3. Data Pre‑processing

Read raw taxi GPS data with pandas, load the study area shapefile with GeoPandas, and then apply built‑in cleaning functions:

<code>import pandas as pd
import geopandas as gpd

data = pd.read_csv('TaxiData-Sample.csv', header=None)
data.columns = ['VehicleNum','time','lon','lat','OpenStatus','Speed']

sz = gpd.read_file(r'sz/sz.shp')

# Remove points outside the study area
data = tbd.clean_outofshape(data, sz, col=['lon','lat'], accuracy=500)

# Remove instantaneous status changes
data = tbd.clean_taxi_status(data, col=['VehicleNum','time','OpenStatus'])</code>

4. Grid Generation

Define a rectangular grid for the study region and obtain grid parameters:

<code>bounds = [113.75, 22.4, 114.62, 22.86]
params = tbd.area_to_params(bounds, accuracy=1000)
</code>

Map each GPS point to a grid cell and aggregate counts:

<code># Assign grid indices
data['LONCOL'], data['LATCOL'] = tbd.GPS_to_grids(data['lon'], data['lat'], params)

# Aggregate vehicle counts per grid cell
grid_agg = data.groupby(['LONCOL','LATCOL'])['VehicleNum'].count().reset_index()
grid_agg['geometry'] = tbd.grid_to_polygon([grid_agg['LONCOL'], grid_agg['LATCOL']], params)
grid_gdf = gpd.GeoDataFrame(grid_agg)
</code>

Visualize the rasterized result with Matplotlib, adding a basemap, colorbar, scale bar, and compass:

<code>import matplotlib.pyplot as plt
fig = plt.figure(1, (8,8), dpi=300)
ax = plt.subplot(111)
sz.plot(ax=ax, edgecolor='k', facecolor=(0,0,0,0.1), linewidths=0.5)
grid_gdf.plot(column='VehicleNum', cmap='autumn_r', ax=ax)
tbd.plotscale(ax, bounds=bounds, textsize=10, compasssize=1, accuracy=2000, rect=[0.06,0.03], zorder=10)
plt.axis('off')
plt.show()
</code>

5. OD Extraction and Aggregation

Extract origin‑destination (OD) points directly from the GPS data:

<code>oddata = tbd.taxigps_to_od(data, col=['VehicleNum','time','Lng','Lat','OpenStatus'])
</code>

Aggregate OD counts on a 2 km × 2 km grid:

<code>params_od = tbd.area_to_params(bounds, accuracy=2000)
od_gdf = tbd.odagg_grid(oddata, params_od)
</code>

Alternatively, aggregate OD counts to administrative polygons:

<code># Without explicit grid parameters (direct spatial join)
od_gdf = tbd.odagg_shape(oddata, sz, round_accuracy=6)

# With grid parameters for faster processing
od_gdf = tbd.odagg_shape(oddata, sz, params=params_od)
</code>

6. Interactive Visualization

Leverage the built‑in Kepler.gl integration for interactive maps. Visualize point density:

<code>tbd.visualization_data(data, col=['lon','lat'], accuracy=1000, height=500)
</code>

Visualize OD flows as arcs:

<code>tbd.visualization_od(oddata, accuracy=2000, height=500)
</code>

Animate individual vehicle trajectories:

<code>tbd.visualization_trip(data, col=['lon','lat','VehicleNum','time'], height=500)
</code>

All visualizations are rendered as interactive web maps that can be explored directly in a Jupyter notebook.