How did your model fare?
My model fared alright, my mse was 7.81 and I would have preferred a lower MSE but all things considered, that is an ok number.
In your estimation is there a particular variable that may improve model performance?
Yes, I think that median income of homeowners could improve model performance because this would give the model a ballpark/range of numbers for the selling price of each house.
Which of the predictions were the most accurate?
Because my data was from Los Angeles, there were outliers for incredibly expensive houses (around 10 million dollars). I think this is why my predictions were most accurate for the lowest priced houses (houses under 1 million), and why there was a sudden drop in accuracy after the 1 million mark.
In which percentile do these most accurate predictions reside?
They reside in the 50th percentile.
Did your model trend towards over or under predicting home values?
It trended towards underpredicting home values.
Which feature appears to be the most significant predictor?
Living space. When I removed it, the mean square error really increased by several thousand. Since larger houses tend to go for more money this makes sense, but bedrooms was a close second.