A DECISION TREE MODEL FOR PREDICTING HOUSEHOLD PRIMARY COOKING ENERGY SOURCES

Abstract

Access to clean and efficient cooking energy sources is a critical concern for sustainable development and public health. In Nigeria, a diverse range of cooking energy sources are used, including biomass, kerosene, electricity, and liquefied petroleum gas. This paper employs a decision tree algorithm to classify the primary cooking energy sources used in Nigerian households. The decision tree model is trained and evaluated using a dataset obtained from the National Bureau of Statistics (NBS) database which was collected through the Nigerian General Household Panel Survey (NGHPS) conducted across various regions in Nigeria. The dataset underwent pre-processing and Variance Threshold was used to select the most relevant features. A dataset representing 4,980 households was used to build the Decision Tree Model. The model was built using Gini Impurity Node-Splitting Criteria together with Pruning using Cost Complexity Pruning Alpha (ccp_alpha) parameter within a range [0.0001, 0.001, 0.01, 0.1, 1, 10]. The study shows that certain household attributes can be used by the model to predict households' Primary Cooking Fuel Sources in Nigeria with an accuracy of 96%, precision of 95%, recall and F1 Score of 93% and 94% respectively. In future, this study can be improved by exploring other Machine Learning Techniques.