Valuation Model using Regression Algorithms for Nepalese Real Estate Marketplace

Abstract

Real estate market is a cornerstone to the Nepalese economy, and is the largest alternative asset class. It is a prime field to apply a systematic model of pricecalculation to sustain the economic growth. The current property valuation process concerns with the mandated minimum price set by the government and the prices of similar nearby properties only. This process does not account for value-based cost increment and personal biases, which further increases the valuation asymmetry in the Nepalese real estate market.The main objective of this study is to address the inefficient valuation process by examining the effects of property features like property area, property location on the price of the property. Having a better understanding of the impact of each property feature on theproperty pricing will provide the stake holders such as property brokers, mortgage lenders, property appraiser and investors better insights in their decision making. This examination uses regression methods and is trained on property size and location as input using 80 percent data split. The model is then tested on unseen data and its performance is compared.The loss function is minimized using gradient descent. The result from this research shows that the property latitude and longitude provided very small sensitivity to the change in property price. The large discrepancies in the property price in the same area for similar property size suggests that more comprehensive models are needed to capture the complex market dynamics of the Nepalese Real Estate Marketplace.