A Python package for the local multiscale analysis of spatial point processes (LomPy)

Abstract

The consideration of multiscale characteristics has emerged as a popular component of geospatial data analysis and modeling. However, the practical implementation of such analysis tasks involves time-consuming and computationally intensive processes that require the integration of knowledge and methods from different disciplines (e.g., quantitative geography, signal processing, and natural sciences) and in which large amounts of data have to be processed. Yet, to date, there is few open-source software that enables an efficient and transparent computational workflow. This paper introduces a Python package for the local multiscale analysis of spatial point processes (LomPy). LomPy is specifically designed for processing and analyzing data that is either irregularly spaced or has large data holes over the spatial territory: a common and methodologically challenging property of geoprocessing operations. The first main function of the package computes the multiresolution quantities, while the second applies them to the extraction of fractal and multifractal features at arbitrarily “local” spatial scales. The third function extends the univariate analysis option to multivariate settings. Powerful tools for regression modeling, density estimation, and visualization are also provided. It should be emphasized that the package is efficient on both vector and raster data structures, ensuring a wide range of applicability in urban data science and beyond.