PREDICTING THE PARISIAN HOUSE PRICE DYNAMICS BEYOND INTEREST RATES WITH HEDONIC PRICE MODEL (HPM) USING MACHINE LEARNING ALGORITHMS

• Main Author: Nur Fadhilah, Salsabila
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/85632
• Institution: Institut Teknologi Bandung

Real estate encompasses a wide variety of activities starting from managing commercial properties and developing land to selling and buying the properties. Understanding the factors that influence the real estate business, particularly in urban locations such as Paris is crucial for making decisions for shareholders. The capacity to accurately estimate real estate prices promotes better decision-making throughout the industry, resulting in a more stable and efficient real estate market. To conduct the quantitative research, the data was gathered from manual data collection from the Paris real estate agency website by gathering 22 entries (7 variables were recorded for each property) with a total 154 observations. This study delves into machine learning algorithms to create comprehensive analysis that analyze factors that influence the value of real estate using Hedonic Pricing Model (HPM). Five factors were examined : location in Paris (code postal), surface (m2), number of rooms and floors, public transport access, real estate types. The results obtained indicate high positive associations of the individual parameters of real estate value: some with surface size and a number of rooms, while access to public transportation showed a slight negative correlation. According to the study’s findings, Ridge Regression is the most effective model by providing reliable predictions with lower error rates than other regression techniques (Random Forest and Lasso Regression) , showing a coefficient of determination (R2) of 0.62 and Root Mean Square Error (RMSE) of 345,512.13. Recommendation for stakeholders includes using price prediction technology thoughtfully, changing risk management tactics, investment optimization methods, and planning for urban projects. In the future, researchers are able to critically understand analysis limitations and combine statistical analysis with human perception and qualitative observations. The objective of this study is to help the stakeholders make an informed decision on real estate with deep analysis by employing a price prediction model.