Digital Injustice: A Case Study of Land Use Classification Using Multisource Data in Nairobi, Kenya

Wenlan Zhang; Faith Taylor, Chen Zhong
International Conference on Geographic Information Science
Academic publication
September 2023


The utilization of big data has emerged as a critical instrument for land use classification and decision-making processes due to its high spatiotemporal accuracy and ability to diminish manual data collection. However, the reliability and feasibility of big data are still controversial, the most important of which is whether it can represent the whole population with justice. The present study incorporates multiple data sources to facilitate land use classification while proving the existence of data bias caused digital injustice. Using Nairobi, Kenya, as a case study and employing a random forest classifier as a benchmark, this research combines satellite imagery, nighttime light images, building footprint, Twitter posts, and street view images. The findings of the land use classification also disclose the presence of data bias resulting from the inadequate coverage of social media and street view data, potentially contributing to injustice in big data-informed decision-making. Strategies to mitigate such digital injustice situations are briefly discussed here, and more in-depth exploration remains for future work.