Predictive Rental Values Model for Low-Income Earners in Slums: the Case of Ijora, Nigeria

Abstract

It is well known most often that values of properties tend to hike at the effluxion of time. This has necessitated the adoption of predictive models in interpreting outcomes in the property market in the future. Earlier studies have been oblivious of such models' outcomes as it affects any focal group, particularly the vulnerable. This present study focuses on the low-income earners found in the slum. The Ijora community in Lagos was the highlight of this study, particularly Ijora Badia and Ijora Oloye, regarded as slums according to the UNDP report. The entire fifty-two (52) local agents in the Ijora community were surveyed in cross-sectional survey research that entailed the questionnaire's issuance. The nexus of data collection, pre-processing, data analysis, algorithm application, and model evaluation resulted in retrieving rental values within the years 2010 and 2019 on two predominant residential property types of self-contain and one-bedroom flats found within the community. Three selected algorithms, Artificial Neural Network (ANN), Support Vector Machine, and Logistic Regression, were essentially used as classifiers but trained to predict the continuous values. These algorithms were implemented through the use of Python's SciKit-learn Library and RapidMiner. The findings revealed that though all three models gave accurate predictions, Logistic Regression was the highest with low error values. It was recommended that Logistic Regression be applied but with much data set of property values of low-income earners over much more period. This study will contribute to the Sustainable development goals(SDG) 11(Sustainable cities and communities) of the United Nations to benefit developing countries, especially in sub-Saharan Africa.