Hyperparameters enable machine learning algorithms to be customized for specific datasets. Choosing the right hyperparameters is a challenge often faced by machine learning practitioners. With this research, tuning of hyperparameters for regression models was explored. Models predicting house prices in King County were created using a detailed suite of regression algorithms. Traditional approaches, and evolutionary algorithms, for improving model accuracy were evaluated. A variety of feature selection methods and hyperparameter tuning using grid search, random search and pipeline optimization were also studied as part of the traditional approaches. Furthermore, evolutionary algorithms were applied to model optimization. In this paper, it is shown that an evolutionary approach, implemented with TPOT, achieves the highest accuracy for a regression model based on the King County dataset. Regarding metrics, combining the RMSE and metrics is shown to be an effective means of determining model accuracy. Finally, greedy feature selection performed best when a variety of feature selection methods are compared.