A Quick Question Concerning Repeated Measures of ANOVA.

#1
Hello Everyone!

I am working on the capstone thesis project for my Master's degree and hoping for some professional assistance to determine if I am heading in the right path. To summarize my project, I am testing to see which resource (CPU, Memory, or Bandwidth) displays the greatest improvement in end-user performance against different DDoS attacks. My data is kind of like this...

data1,2,3 = baseline_cpu, baseline_memory, baseline_bandwidth + attack1x5; then attack2x5; then attack 3x5
data4,5,6 = increased_cpu, baseline_memory, baseline_bandwidth + attack1x5; then attack2x5; then attack3x5
data 7,8,9 = baseline_cpu, increased_memory, baseline_bandwidth +attack1x5; then attack2x5; then attack3x5
data 10,11,12 = baseline_cpu, baseline_memory, increased_bandwidth +attack1x5; then attack2x5; then attack3x5

I don't know if it matters for choosing a statistical method, but each attack was conducted for a 5 minute duration. So, all together, the important information is what attack is occurring and what resource is increased at what time so that I can see which, if any, increased resource did display improved performance from the end-user perspective.

For the statistics I was thinking of using ANOVA, but just realized that repeated measures of ANOVA may be the better options. I have taken two statistics course so far, so I don't have a lot of experience with this, but hopefully enough to get me through this. I am really interested to see what statistical method would be the best choice! :D

Also, are there any reputable free tools that I can find online to help with the mathematics of this?

Thanks a lot for your assistance.
-Cyanide


Edit: Since posting this I have spent some time reading through my Statistics book from a few semesters back. From this I think that the appropriate method would be to use Factorial ANOVA for this. I am thinking this because I have two groups (Resource + Attack) with multiple factors (Level of Resource + Method of Attack). Let me know if anyone has advice concerning any other Statistical methods that may be useful.
 
Last edited:

noetsi

Fortran must die
#2
If you use a commerical software you won't need to do the math to do the calculations. Generally speaking, outside statistics journals anyway, few if any show the math involved in ANOVA. They simply report the results (and of course the theory, the lit review, how you operationalized your data etc). As a starting point you should look up some articles on repeated subject ANOVA and see how they report the results.

I am not an expert on repeated subject ANOVA although I have taken it in a class. I am not sure I understand your data, but repeated measures usually works as an initial measurement, an intervention, a new measurement, an other intervention etc. Its common when you do this to have a control group that does not go through the interventions. I think your data supports that, although I am not certain.