Machine learning applications for urban geospatial analysis: A review of urban and environmental studies

The application of machine learning (ML) in geospatial analysis has witnessed a significant upsurge, particularly in the last five years. This surge is linked to exponential developments in artificial intelligence (AI) technologies and the extensive demand for their applications in geospatial analysis fields such as urban and environmental studies and planning. Given these rapid developments, understanding the capabilities and applications of ML remains an understudied but crucial area of research. This study aims to narrow this knowledge gap by conducting a systematic literature review of ML in geospatial analysis. This review identifies the thematic area, ML technique and domain, technique, and data type of each reviewed piece of literature, thereby providing insights into the progression and trends of ML in geospatial analysis. ML techniques are categorised into two main components: ML and deep learning, while two key application domains are identified: computer vision and natural language processing . It also identifies six primary application areas: classification, detection, extraction, clustering, regression, and optimisation. Moreover, this study offers an in-depth investigation of state-of-the-art ML applications across geospatial analysis areas, mapping the forefront of ML advancements in each category, thus providing a detailed picture of the present status and future direction of ML in geospatial analysis. In addition to mapping the current landscape of ML in the geospatial sciences, it anticipates future developments, paving the way for novel advancements in spatial data interpretation and applications that inform urban and environmental planning research and practice.