cameronshogg / syde312

SYDE 312 Applied Project - Linear Regression to Predict Housing Prices in Melbourne, with OLS and L2 Regularization

SYDE 312 Applied Project

Linear Regression to Predict Housing Prices in Melbourne, with OLS and L2 Regularization

Abstract:

This report details an attempt to construct a Multiple Linear Regression model to predict housing prices in Melbourne based on several factors, including number of bedrooms, bathrooms, and distance of the property from the downtown core. Initial pre-processing was completed in Python to select important features and remove null rows from the dataset. An Ordinary Least Squares model was then computed in Matlab using QR Decomposition, and a Ridge Regression model was fit using the Normal Equations. The Ridge Regression model performed better, but both models had high Root-Mean-Square Error, meaning that neither predicted new observations well.