Thesis (MSc) - Cyprus International University. Institute of Graduate Studies and Research Management Information System

This research aims to improve the market stability of Northern Cyprus and consumer trust through property recommendations based on the achieved high-accuracy price predictions. Using a dataset collected between 2023 and 2024, captured from 101Evler.com using Python, a leading property search portal based in Northern Cyprus. The dataset includes data on 430 residential properties on the island, the study explores the unique dynamics of Northern Cyprus characteristics as an unrecognized state offering lower property prices and investment opportunities than other countries around the world. Machine Learning algorithms were used such as Random Forest, Gradient Boosting, and XGBoost, compared their results and their performance was evaluated by using R-squared (R²) and Mean Absolute Error (MAE) metrics. The results proved that Random Forest Regression outperformed the other models, achieving the highest R² scores and the lowest MAE values, the properties recommended were based on the results of the high predicted values rather than the actual prices which illustrate the potential of being overpriced in other platforms in Northern Cyprus. Our research applied the property price prediction analysis methodology and utilized the Dataiku platform and Python for model development and recommendation. Our findings highlight the potential of recommending worth buying properties, enabling more informed decisions through accurate price predictions.