Hello all,
I am trying to do an agreement analysis to verify how similar are the time-series measurements taken by two devices. Basically I have 2 curves representing values measured over time with each device, and I want to say how similar these measurements are.
I have other metrics in my analysis, but I was looking into CMC (Kadaba, 1989) to be a global metric. I know it is often used in gait analysis literature for reliability analysis, where curves taken by the same measurement device, but in different days, are compared. This coefficient represents similarity between two curves, so I was considering using it as a metric of agreement between the two time-series measurements I have, one from each device. I was wondering if there is any statistical assumption behind CMC that prevents me from doing that, I couldn't find much about it.
Thank you!
Ana Christine de Oliveira An analysis of variance (ANOVA) can be used to measure similarities and differences. Also, multiple correlations can be used.
Please see this article
Article Method agreement analysis: A review of correct methodology
Dear Mohamed T. A. Abdelmawgoud ,
Thank you very much for taking the time to answer. I believe ANOVA wouldn't apply for a time-series dataset, since data points are not independent from each other, which I believe is one of the assumptions of the analysis. I think that would be definitely an option if I could summarize the entire time-series curve into a single metric per subject, but it's not what I want to do here.
Thank you for your suggestions!
Dear Pervaiz Iqbal ,
Thank you very much for sharing the paper! It seems to be very applicable to this case. I was trying to find something like this, but I guess I was using the wrong terms in my search, so I really appreciate it!
Completely in agreement with the magnificent answer of the also magnificent researcher Dr. Pervaiz Iqbal; curiously I was going to hang something similar.
- Badges
- Science topic