Skip to main content
Cross Validated

Return to Answer

deleted 27 characters in body
Source Link
ely
ely
  • 2.4k
  • 19
  • 32

One option is to plot additional curves from the green data. Here you plot one solid green curve connecting the mean values for each time to first detection.

Instead, you could compute the 25th percentile of all the values at a given time to first decision. Repeat that for each time to first detection, and you'll get a whole curve of 25th percentile values. Do that again for 75th percentiles. Now plot those as, perhaps, dashed light green curves, which will bracket the solid green line.

Readers can then visually compare the red line not just to the solid green line, but also to the distribution of values spreading between the 25th percentile curve and the 75th percentile curve.

Swapping the solid lines away from mean values and toUsing median values mightfor the solid curves could also be good. And, of course, instead of using the quartile values (25th, 50th, and 75th percentiles at each time to first detection) you can use the mean and +/- one standard deviation at that point. I think ordinal stats are a bit more informative, especially if there are outliers, but it could work.

One option is to plot additional curves from the green data. Here you plot one solid green curve connecting the mean values for each time to first detection.

Instead, you could compute the 25th percentile of all the values at a given time to first decision. Repeat that for each time to first detection, and you'll get a whole curve of 25th percentile values. Do that again for 75th percentiles. Now plot those as, perhaps, dashed light green curves, which will bracket the solid green line.

Readers can then visually compare the red line not just to the solid green line, but also to the distribution of values spreading between the 25th percentile curve and the 75th percentile curve.

Swapping the solid lines away from mean values and to median values might also be good. And, of course, instead of using the quartile values (25th, 50th, and 75th percentiles at each time to first detection) you can use the mean and +/- one standard deviation at that point. I think ordinal stats are a bit more informative, especially if there are outliers, but it could work.

One option is to plot additional curves from the green data. Here you plot one solid green curve connecting the mean values for each time to first detection.

Instead, you could compute the 25th percentile of all the values at a given time to first decision. Repeat that for each time to first detection, and you'll get a whole curve of 25th percentile values. Do that again for 75th percentiles. Now plot those as, perhaps, dashed light green curves, which will bracket the solid green line.

Readers can then visually compare the red line not just to the solid green line, but also to the distribution of values spreading between the 25th percentile curve and the 75th percentile curve.

Using median values for the solid curves could also be good. And, of course, instead of using the quartile values (25th, 50th, and 75th percentiles at each time to first detection) you can use the mean and +/- one standard deviation at that point. I think ordinal stats are a bit more informative, especially if there are outliers, but it could work.

Source Link
ely
ely
  • 2.4k
  • 19
  • 32

One option is to plot additional curves from the green data. Here you plot one solid green curve connecting the mean values for each time to first detection.

Instead, you could compute the 25th percentile of all the values at a given time to first decision. Repeat that for each time to first detection, and you'll get a whole curve of 25th percentile values. Do that again for 75th percentiles. Now plot those as, perhaps, dashed light green curves, which will bracket the solid green line.

Readers can then visually compare the red line not just to the solid green line, but also to the distribution of values spreading between the 25th percentile curve and the 75th percentile curve.

Swapping the solid lines away from mean values and to median values might also be good. And, of course, instead of using the quartile values (25th, 50th, and 75th percentiles at each time to first detection) you can use the mean and +/- one standard deviation at that point. I think ordinal stats are a bit more informative, especially if there are outliers, but it could work.