Informal settlements are a common phenomenon in urban development in Indonesia, where societies inhabit a neglected, unused, and unsupervised area, including floodplains. The communities often rely more directly on the provision of ecosystem services derived from urban rivers. Although the physical characteristics are visible, there is still a challenge to map and predict the pattern of informal settlements in urban development. This paper offers a critical review of the use of machine learning in land change modelling as a tool to simulate the existing pattern of informal settlements along the Ci Kapundung River and its surrounding area in Bandung, Indonesia. The model uses an explanatory variable test procedure to assess how significant environmental drivers and urbanization (e.g., slope, elevation, proximity to the river, roads, the distance from existing development, and the likelihood of land cover change) are in the generation process of informal settlements. This study shows that human decisions and the ability of inhabitants to adapt to the new environment have strongly affected the pattern, which cannot be shown from the outcomes but are potentially reflected when using agent-based models.
