Real-Estate Values Prediction using Machine Learning Techniques

Abstract

Real-Estate is one of the important businesses in Pakistan which helps the country to boost its economy. The real-estate business is playing a vital role in economic growth and stability in economic conditions worldwide. The value prediction is one main aspect to make investments in this business. Real estate is contributing more than 9% of Pakistan’s GDP and the market capitalization of real estate is over $1 trillion by the end year 2020. Our research is based on the value prediction of real estate by applying four different Machine Learning models to two different datasets. The framework proposed in this study is mainly consists of four steps, step I is data acquisition, step II is data pre-processing, step III is exploratory data analysis and Step IV is dimensionality reduction, to find out key factors that affect the market value of the real estate. Two datasets are used for experimentation and model validations namely KCUSA dataset and zameen.com dataset of Pakistan region. In our research, we used Multiple Linear Regression, Random Forest, Gradient Boosting Regression, and Keras Regression for real estate values prediction and compare the performance of these models. Among all these models Random Forest produced excellent results by establishing a strong relationship between attributes of both datasets.