The Application of Partial Least Square Regression to Develop a Model for Reliable Investment Valuation Estimates in South-East Nigeria

Abstract

Against the increasing criticism of the reliability of the conventional investment method of property valuation practice in Nigeria, this study examine the ability of the Estate Surveyors and Valuers to estimate accurately the selling prices of residential properties in South-East Geopolitical Zone of the country with a view to formulate a model for reliable value estimates. Questionnaires were administered in a sample of randomly selected forty (40) Estate Surveyors and Valuers in the four selected towns of (Aba, Enugu, Onitsha and Owerri) in the South-East Geopolitical Zone. From the analysis of variance, the total of degree of freedom is seventeen, while the sum of square is 2682.44 and P-value in the table is 0.00, meaning that there is a statistically significant relationship between the Prices and Valuer estimates. The study further used the application of partial least square (PLS) regression model to develop model for perfect reliable price estimates. The model is thus:

P=2.250.13_v10.23_v20.08_v30.17_v40.17_v5+0.1_v60.03_v70.25_v8+0.14_v90.02_v10+0.07_v11+0.26_v12+0.37_v13+0.07_v14+0.07_v15+0.09_v16+0.1_v17+0.06_v18+0.08_v19+0.2_v20+0.2_v21. The model developed has the highest independent (X) variance, lowest level of the predicted error and highest coefficient of determination (R-square) between the value estimates and market prices and gives optimal and reliable result.

Keywords: Partial least square, regression analysis, investment method of valuation, Valuation reliability, South-East Zone of Nigeria.