Because of this I have a heavily clustered correlation scatter plot which makes it difficult to spot any direction of the regression line. However, there is a correlation of about 0.5. When running the regression I get extremely small values for the beta coefficients which makes interpretation difficult. What would be a good way to run a regression with this data so that I can get some more plausible coefficients?