No its not a good reasonable fit to use a linear regression line. Why not? Because r = 0.15 is closer to r = 0 than it is to r = 1. This is fairly weak linear correlation. If r was say something like r = 0.7 or higher, then it would be stronger linear correlation.
