A Recommendation Engine to Estimate Housing Values in Real Estate Property Market

Abstract

A recommendation engine is an automated machine learning technique developed to filter some entities called people, properties or objects, products and movies based on user preference. This can be employed to express recommendation based on user interest determined by its market value. The increasing demand and limited in supply of housing properties limits the existing methods of operation to bear a heavy burden in recommending housing properties. This leads to information overload which made it difficult to search and recommend properties with respect to user's specific interest. These can be affected by its location and condition of property. Therefore; we proposed the use of a recommendation engine with collaborative technique in providing optimal solution to the challenges facing the housing market. This work is aimed at developing an efficient recommendation engine to estimate housing values. This serves to overcome the problem of information overload, help understand what the user wants and recommend properties with the most popular features. The implementation was done using logistic regression and K-nearest neighbor techniques with Python. The performance was improved with some fine-turned hyper-parameter values using the kaggle online dataset. The K-nearest neighbor produced 100% prediction accuracy recorded to be better than the logistic regression with 54.6% accuracy rate.