Abstract:
autoESDA is a Python library developed with the aim of automating the Exploratory Spatial Data Analysis (ESDA) process. This is done by generating a HTML report made up of various ESDA graphs and statistics calculated according to the input dataset, requiring no other inputs from the user. ESDA (local spatial autocorrelation specifically) in Python has been a challenge for raster datasets, with software support lagging behind alternative platforms such as R. This dissertation documents the improvements made to the original library. These improvements include the support for raster datasets, an updated architectural design, and other minor, cosmetic improvements. The performance of the updated version of autoESDA is evaluated by investigating how its processing time varies according to vector and raster datasets that differ in size and complexity. These results are then discussed as a measure of how well the library has achieved its goal of automating the ESDA process. Finally, a roadmap for further improvements to the library is discussed.
