• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Data Interpretation

Q

QueHuong

Platinum Member
In the following pictures is a graph of data points. You can see from the graph that as you go along the X-axis, the data points get slightly closer together. What it represents is that as I perform more trials (X-axis), the smaller the impact of those trials (Y-axis)...ie, I'm reaching the point of diminishing returns. Other than just eyeballing the graph, is there a statistics tool out there that can help me quantify why I should only be performing a certain number of trials?

For example, just from the looks of it, after the 6th trial, the differences compared to the previous data point is starting to be negligible. But what math tools can I use to show 6 (or some other number) is the minimum number of trials I should do, where more trials would just simply be not worth the time it takes. Thanks.


Graph
 
going to hazard a guess here and first of all state that the y-values dont saturate at the 6th point: they're still going up.

i'd do a best fit and fit and see which function best fits the plot, but to do that I think you'll want your trials on the x-axis and data values on the Y.
 
Auggie: My mistake...I worded the first post wrong. The graph is correct though: the data points are on the Y-axis and the number of trials performed are on the X.
 
i imagine you could set up some sort of confidence interval and when the mean Y value starts falling within that interval you have achieved a minimum "n" value (# of trials)

I have a bunch of stat textbooks that teach solving for n= ? based on such things (but they are home).

Do you have access to any stat books?
 
Do you have a truth estimate for the y-axis? If not, it sounds like you're just looking to see where the differences between N, N+1 go to zero. You may be better off plotting the one step differences (~derivative) to see where this curve goes to zero.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top