INFO_7390_Group(Xiangyu Chen, Sike Dong, Fanyu Mou)
House is one of the most important part for family and plays important role in economic development. Therefore, this project is aimed at analyzing contributions of US housing prices, then predicting. For our methods, we use Linear Regression and Deep Learning Regression to analyze the factors and their relationships with housing prices, then applying Random Forest to do the prediction. In result, we find out several main factors like the amount of rooms and living square feet. The results show that Random Forest and Linear Regression can provide better predictions.
As an important source of economic development, real estate has become a pillar industry for the United States. Conor Sen, the investment manager of New River Invest even said that Real Estate would become the economic engine in the next 5 years. Buying a house is a crucial and most expensive decision for each family. Therefore, our project is to go insight of this industry, figuring out the factors for US housing price and to predict, we have 4 datasets for four different areas: King County, Iowa, San Luis and California.
Correlation Matrix to find out the key contributors toward housing price; Multiple Regressions(6 kinds) and Deep Learning Regression to analyze the relationships between key factors and housing price; Random Forest to predict housing prices.
Since one of our goal is to analyze the relationships among housing price and key factors, combining with all the results we get for the datasets, Linear Regression is the fittest one.
In fact, based on economic theory and our analysis overall, the most important factors that would affect the housing price are the economic development, policies and demands etc. The factor like size, quality or number of rooms etc, play but not the decisive roles.