Geospatial and socioeconomic prediction of value-driven clean cooking uptake

• Published in: Renewable & sustainable energy reviews 2024-03, Vol.192, p.114199, Article 114199
• Main Authors: Flores Lanza, Micaela, Leonard, Alycia, Hirmer, Stephanie
• Format: Article
• Language: English
• Subjects: Clean cooking; LMICs; Machine learning; QGIS; User-perceived values; Value perception
• ISSN: 1364-0321
• DOI: 10.1016/j.rser.2023.114199

Understanding the community-specific values and needs of consumers is essential for effective targeting and planning of energy services such as clean cooking. Many clean cooking programmes do not however consider these values and needs in targeting, as they can be difficult and time-consuming to ascertain. This work therefore explores whether community needs and values related to cooking can be predicted, using a novel approach that understands the relationship between socioeconomic, demographic, and geospatial data. Specifically, this study investigates (i) which values are most closely linked to cookstoves in rural Uganda; and (ii) whether it is possible to predict cookstove prioritisation and related values using openly-available data. Using machine-learning approaches, user-perceived value data from 199 rural low-income households in Uganda are mapped against socioeconomic, demographic, and geospatial data to identify correlations and intersections. The values most closely related to cookstoves were found to be food security, time benefit, accessibility to services, fixed costs, and being healthy. The most important parameters in predicting who would hold these values were found to be: the number of people living in a house; age; quintile 2 of the wealth index; annual accumulated precipitation; forest density; night time luminance; and distance to water source, nearest forest within ten kilometers, and nearest road. This study takes a first step towards enabling energy service providers to target areas with a greater likelihood of uptake based on open-source datasets. While cooking in Uganda is analysed herein, the proposed method can be applied for different geographies and energy services. [Display omitted] •We analyse whether cooking-related needs/values can be predicted with open data.•Geospatial, demographic, and socioeconomic data are mapped against values.•A dataset from 199 households in rural Uganda is used as a case study.•Top cookstove-related values are food security, time benefit, and service access.•Household size, age, wealth, precipitation, and forest density are determinants.