House-Price-Prediction

The process of valuing and marketing a house is challenging. Traditional methods used by realtors only consider a limited number of parameters over a relatively small sample size, which can lead to houses being misvalued. There is also the potential for realtors to have an unconscious bias when valuing a property, which can be detrimental to prospective buyers. These shortcomings of traditional valuing techniques have presented the need for more advanced, accurate, and efficient methods of valuing a house. The purpose of this project was to take part in the House Prices- Advanced Regression Techniques competition. This competition involved the analysis of 79 different variables describing numerous aspects of houses in Ames, Iowa, and subsequently developing machine-learning models capable of predicting the final prices. This notebook provides a detailed analysis of the data and the experimentation of three different machine- learning models, namely Linear, Ridge, and Lasso Regression, to investigate their prediction accuracy. The effect of various parameters on each model's accuracy was investigated, and different machine-learning tuning methods were explored.