Prediction of the likelihood of households food security in the lake victoria region of Kenya

Abstract

This paper considers the modeling and prediction of households food security status using a sample of households in the Lake Victoria region of Kenya. A priori expected food security factors and their measurements are given. A binary logistic regression model derived was fitted to thirteen priori expected factors. Analysis of the marginal effects revealed that effecting the use of the seven significant determinants: farmland size, per capita aggregate production, household size, gender of household head, use of fertilizer, use of pesticide/herbicide and education of household head, increase the likelihood of a household being food secure. Finally, interpretations of predicted conditional probabilities, following improvement of significant determinants, are given.