PyNKDV: An Efficient Network Kernel Density Visualization Library for Geospatial Analytic Systems

Tsz Nam Chan, Rui Zang, Pak Lon Ip, Leong Hou U, Jianliang Xu

Abstract: Network kernel density visualization (NKDV) is an important tool for many application domains, including criminology and transportation science. However, all existing software tools, e.g., SANET (a plug-in for QGIS and ArcGIS) and spNetwork (an R package), adopt the naïve implementation of NKDV, which does not scale to large-scale location datasets and high-resolution sizes. To overcome this issue, we develop the first python library, called PyNKDV, which adopts our complexity-reduced solution and its parallel implementation to significantly improve the efficiency for generating NKDV. Moreover, PyNKDV is also user friendly (with four lines of python code) and can support commonly used geospatial analytic systems (e.g., QGIS and ArcGIS). In this demonstration, we will use three large-scale location datasets (up to 7.71 million data points), provide different python scripts (in the Jupyter Notebook), and install existing software tools (i.e., SANET and spNetwork) for participants to (1) explore different functionalities of our PyNKDV library and (2) compare its practical efficiency with existing software tools.