| 68 | | 12. In the DREAM parameter 2d correlation plots are very good but don’t actually quantify how large the correlation is. It might be useful to have some scale markers on the axes. The 2d plots are of course very small, so normal axis labels would likely look a mess, perhaps try sticks on the sides at 0 and say +-3sigma. (Normal correlation coefficients as I recall say how many standard deviations parameter A moves if parameter B moves one standard deviation.) |
| 69 | | |
| 70 | | PAK: yes, I left off the labels because the plots would be too busy. Since the points used on the correlations are the same 95% interval as used on the 1-D variable plots, you can use those plots to read off the actual values, or look on the console log to get a parameter table. I created a ticket to print the correlation coefficient on the 2-D plots. |
| | 68 | 12. In the DREAM parameter 2d correlation plots are very good but don’t actually quantify how large the correlation is. It might be useful to have some scale markers on the axes. The 2d plots are of course very small, so normal axis labels would likely look a mess, perhaps try sticks on the sides at 0 and say +-3sigma. (Normal correlation coefficients as I recall say how many standard deviations parameter A moves if parameter B moves one standard deviation.) [PAK: yes, I left off the labels because the plots would be too busy. Since the points used on the correlations are the same 95% interval as used on the 1-D variable plots, you can use those plots to read off the actual values, or look on the console log to get a parameter table. I created a ticket to print the correlation coefficient on the 2-D plots.] |
