House Price Forecasting by Implementing Machine Learning Algorithms: A Comparative Study

Discerning property value via state-of-the-art machine learning techniques can evolve the current real-estate market and expose it to the technological frontiers of the modern world. This can potentially have sanguine domino effects such as opening the market to new investors as a result of technically backed price values. The current research paper strives to capitalize on this opportunity by analyzing information and data from an existing online marketplace for buyers and sellers in this industry. It is conjectured that precise prediction of house prices in a particular location through data analytics will create a candid market where prices are not arbitrary, ensuring openness in the market through P2P opportunities which will eliminate middleman charges. The research ventures to extrapolate machine learning techniques to create a model that predicts house prices in Bangalore using a plethora of algorithms such as linear regression, bagging classifier, K-nearest neighbour, XGB, decision tree, gradient boosting, and random forest. An incremental approach is deployed to gather and streamline data, clean, visualize, model and evaluate the models produced. The research is completed with a result from the comparative study, showing the most appropriate algorithm for the given data available is the random forest algorithm.