Ames Housing Price Prediction

Summary of findings

Model comparison

Ridge regression by far has the best test score yet at 88.82%. With 31 features now, it is now more ideal and easy to explain to relevant stakeholders. Ideally we would always want to have a model that is complex enough to fit all the points but still having enough simplicity to explain your model.

Additionally, the Ridge Regression model does not look to be overfitted as they perform better on the test score.

Top 5 features from final model

The top feature that highly influences the price at a coefficient of 48313.86 is a property located at Stone Brooks neighbourhood. It is no wonder the best predictor as this neighbourhood is very convenient and is the closest to Iowa State University, the largest university in Iowa state. [1] It is also very close to downtown and is just about a 10 minute drive away. The other two neighbourhoods, Northridge Heights and Northridge are situated close together and is slightly further away to then University but is equidistant to downtown as compared to Stone Brook neighbourhood. Moreover all three neighbourhoods are very close to elementary, middle and high schools and have amenities really close by. [2]

However, it is interesting to note that the land contour that is hilly is valued more than the other land contour features. It is perhaps has a good privacy rating among home dwellers in Ames.

At number five, Overall Quality of the property takes the spot and is no surprise that it is one of the top predictors in property prices, though the coefficient is much lower than being in the top three neighbourhoods.

Therefore, we can conclude that having a property in those top three neighbourhoods is way more important in highly influencing property prices than Overall Quality.

For more info, you can visit more on my Github.