HYBRID MACHINE LEARNING MODEL FOR HOUSING PRICE PREDICTION USING FEATURE SELECTION
With the exponential rise in the population, the real estate sector is seeing massive growth. Technology can be mass-produced, and resources can be replenished, but the inhabitable land is finite and will always be. Strangely, even after the technological revolution that we are witnessing today, real-estate prices are still determined manually, based on factors like the per-square-meter rate of the last sales deed in that locality, elegance of exteriors and interiors, quality of fittings and a general human-sense of beauty and aesthetics. This ambiguity in rates is hitting all the stakeholders, including the buyers, sellers and property dealers. With no formulas in place, the prediction of a house’s price based on images and other data points is still a challenge. This paper makes a successful attempt to solve this challenge using machine learning and image processing. The paper proposes a way to predict the valuation of a house accurately, based on its images and other relevant data. The proposed process involves training machine learning to use exterior and interior images. Later, selecting suitable features using the GA-reinforcement boost strategy. Finally, applying the above to the XG-Boost Regressor to generate results. This algorithm, if successfully deployed, will be beneficial to both sellers and buyers, because it sets a data-based benchmarking for evaluating the property.
GA-reinforcement strategy, XG boost Regressor, Machine-learning models, real estate valuation