Apartment Purchase Suitability Prediction Using Explainable Machine Learning

Abstract

Apartment purchase suitability checking is important for customers due to apartment price, location, and customer's income issues. Existing research works could not incorporate essential features and ML models for apartment purchase suitability checking. The accuracy of their work is also low. They did not investigate proper feature selection method, data preparation, class imbalance issue solving, and hyper parameter tuning techniques. This paper introduces a ML based framework for predicting the suitability of apartment purchases by considering a number of new factors, advanced preprocessing, feature selection, and hyper parameter tuning methods. This work’s preprocessing procedures included class balancing, addressing missing values, normalization, and ensemble feature selection technique. This paper examined 6 ML models for best prediction model selection and Grid Search CV is used for hyper parameter tuning. The random forest appears as most suitable ML model with 89.48% accuracy and 89.50% F1 score. According to the results, the proposed RF model outperformed previous works by offering more than 2 percent gain in accuracy and F1 score. This paper also included feature importance analysis based on SHAP.

Received: 10 December 2025 | Revised: 17 March 2026 | Accepted: 9 May 2026

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support this work are available upon reasonable request to the corresponding author.

Author Contribution Statement

Asim Foize Aimon: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Visualization. Mahfuzulhoq Chowdhury: Conceptualization, Methodology, Investigation, Resources, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration.