I measured 20 times the runtime of three different C++ programs and I've got the following results (average):
runtime_program_1 = 26,43 ms;
runtime_program_2 = 27,02 ms;
runtime_program_3 = 13,56 ms.
Qualitatively, program 3 is faster than the other two, but how can I use quantitative analysis to support this statement? i.e., is program 3 significantly faster than programs 1 and 2? Also, is program 1 significantly faster than program 3? Thank you very much!
Dear Erick
You can go for ANOVA (Treat = 3; Replication = 20), as you would like to test three means. Based on the analysis of 20 observations for the three program, you can conclude the fastness of the program.
Regards
rajiv
I agree with Rajiv. First an omnibus ANOVA, and then follow-up tests (for 1 vs. 3 you can just use a simple t-test; for 3 vs. {1,2} you probably want to set up custom contrast coefficients).
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics