Figure 8: Number of samples required for naïve fit for the “Google” data set. Plots show fit standard error for fits increasing random samples, and a horizontal line is given at the RMSE obtained for all points.