Approaching Accurate Property Valuation Model

Abstract

The need for convenient, effective and accurate prediction models is indispensable for the user to have a quick, easy and more objective process to arrive at the opinion of value. Residential property valuation have economic consideration for arange of stakeholders. Taxation offices utilize property valuations to establish appropriate property tax levels and private investors for purchase purposes. They are also employed by banks to assist with mortgage lending and by insurance agencies to help with risk assessment. Moreover, valuation accuracy has implications for thcountry’sry economy, tax policies and fluctuation in interest rates. Account the importance, property valuation requires impartiality and technicality for a transparent market to avoid uncertainties. In Pakistan real estate valuers are using traditional practices for property valuation. Assessors of real estate frequently utilise the comparable sales approach, whose primary idea is to determine the worth of a property by evaluating and contrasting the costs of homes that are similar to it and are typically situated close by. Traditional techniques are mired with inaccuracy and errors. Which is leading to creation of significant deviations between actual and assessed market values of properties .A poorly designed valuation system will under asses the value of taxable base and reduce the revenue potential. In Pakistan, there is a lack of research on property valuation and there has been no coordinated attempt to model the Pakistan property market effectively using advanced valuation techniques. Inaccurate property appraisal is a problem that concerns not only Pakistan but the rest of the world and has drawn the attention of academics from all over. Despite its shortcomings, the HPM method for property assessment has received widespread acceptability among real estate scholars, as indicated by past studies. Innovative strategies are required to enhance the accuracy of property valuations. Variety of factors affects property valuation so adopting various Artificial Intelligence (AI) techniques offers advantages i.e. : to effectively analyse big data, find non-linear links between property characteristics, improve housing price forecasting, and greatly aid in the accurate estimation of real estate value. To overcome the shortcoming of HPM approach and other AI based techniques this study develop prediction models for Pakistan. Two modelling techniques, i.e. HPM and Multi expression programming (MEP), a new AI-based technique for efficient Property valuation were introduced in property valuation and is applied in this study. The superiority of MEP viii is established in the literature and is employed because of its ability of the development of practical mathematical expression and predictive capability. Unlike MEP other AI algorithms i.e artificial neural network (ANN), Random forest etc fails to produce a practical mathematical expression. The research objectives are; to identify valuation attributes affecting residential housing prices in Hayatabad Peshawar and rank them based on stakeholders opinion; Development of HPM model for residential property market of Peshawar; Development of MEP model for residential property market of Peshawar; Evaluation of predictive accuracy of both HPM and the MEP model. Study area for this research is Hayatabad, a township in Pakistan sixth largest city Peshawar, province KPK. Important property attributes were identified from literature review and were presented to the relevant stakeholders to rank according to Mean Score (MS). In addition, sales transaction of houses from real estate firms operating in Hayatabad,Peshawar were collected. They were used for the development of HPM and MEP modelfor Hayatabad, Peshawar property market.HPM and MEP model were developed from the same data for comparing the predictive accuracy of both models. The data was divided into two parts: training and testing data points. The results from HPM analysis were unsatisfactory compared to MEP model. The results of the HPM study, which produced a coefficient of determination (R2) value of 0.87, were unsatisfactory. Considering that the mean absolute percentage error (MAPE) score having higher value, 17.2 percent, this also is not sign of accurate property prices. Based on the measures mean absolute error (MAE) and root mean squared error (RMSE), the HPM model are having higher values compared to MEP. The MEP model generates an R2 value of 10.5% and an MAE value of 10.5%, both of which are reasonably accurate property values. Additionally, the RMSE and MAE values were superior to the HPM technique. This suggests that, in terms of property valuation, the MEP model is a more effective replacement for the HPM method. Additionally, when compared to the HPM findings, a large percentage of the MEP's predicted property values had acceptable prediction errors, according to international guidelines. The property valuation in Pakistan is done through traditional methods and not proper valuation practices are adopted yet. The MEP model can be applied as tool for property valuation practices. In addition, this research has provide justification for the incorporating AI based models in both theory and practice for property prediction. MEP model is efficient and effective as it is accurate and provide mathematical expression for property valuation.