Wali Ullah

Volume 1 Issue 1 | Dec 2024

DOI:

Views: 7

Total Downloads: 1

Download PDF

---

Abstract

Regression algorithms play a pivotal role in predictive analytics in data mining and artificial intelligence (AI). Predictive analytics, including property price forecasting, is important to numerous stakeholders, including buyers, sellers, businesses, and government agencies, as it facilitates informed decision-making. This study uses a comprehensive set of machine learning models, including standalone and ensemble models, which are not considered in existing studies using two different datasets. We trained multiple machine learning models on two different datasets of variable nature to obtain and learn a more accurate function that maps the values of dependent variables to the dependent variable, i.e., price, and finally identify the optimal price prediction models that perform well on both datasets. Furthermore, in this research work, we attempt to evaluate the performance of seven different machine learning algorithms, namely linear regression, lasso regression, ridge regression, decision tree regression, machine vector machines, random forest regressor, and gradient boosted regressor, by using four different evaluation metrics on two different datasets. Our results show that in dataset 1, the gradient-boosted regressor outperforms its counterparts and has excellent prediction accuracy, while in dataset 2, the linear regressor outperforms others. 6-fold cross-validation was used to obtain more reliable and valid evaluation scores. Overall, the study found that the gradient-boosted regressor (GBR) is the preferred model for price prediction, although the linear regressor showed slightly better performance in dataset 2, which is negligible.
Keywords: Price Prediction, Linear Regression, Random Forest, Gradient Boosting, SVR, Machine Learning, Predictive Analytics