Visualization of Longitudinal Data by Fitness Orbits: Application to Data on Rural Households at Agincourt

Charles Lebon Mberi Kimpolo, University of the Witwatersrand
David Sherwell, University of the Witwatersrand
Mark Collinson, University of the Witwatersrand

In this paper we present first demographic results by visualization of fitness orbits using a new mathematical modeling technique. We use three demographic variables namely biological mother's out-migration, household head is a minor and adult death to investigate their impact on children's educational progress. A sample of 2669 Agincourt households was investigated from 1998 to 2007. We define educational default by those scholars who have failed more than three years in their school life. The sample population then splits into favorable (45.83%) and unfavorable (54.17%) sub-populations. Automatic geometric clustering of fitness orbits shows that 0.045% of Agincourt households were headed by minors and dominant dynamics were related to mother's out-migration followed by adult death. We find that the defaulting population was dominated by mother's out-migration while the non-defaulting population was characterized by mother staying home. These results suggest that Agincourt child's educational default is related to biological mother out-migration.