3
$\begingroup$

I have two sets of independent data including 10 samples.

It's like,

      set A         set B
1     961             965
2     919             934
3     886             921
4     896             913
5     881             967
6     918             924
7     927             925
8     942             915
9     477             896
10    921             930

These are the time taken to arrive destinations for vehicles.

Because the environment is dynamic, it sometimes arrives early and late.

And, set A is the results from transportation policy A, and set B is from transportation policy B.

I assume that I don't know whether policy A and policy B are same or not.

And, What I want to know policy A is different from policy B? and how different?.

Before thinking about statistics, my method to calculate was just, min value of set A < average of set B < max value of set A.

So, if the average is in range of above inequation, policy B is different from policy A.

To be accurate, I considered Jensen-Shannon divergence (because it should be symmetric).

But I'm not sure that Jensen-Shannon divergence is the best though the samples are few.

How can I compare two sets?

Improve this question
$\endgroup$
6
  • $\begingroup$ We cannot answer questions about "comparing data", without context, because statistics is about answering applied questions, real-life questions, so we must know about the real context that produced your data, and what question, real questions, you want to use the data to answer. Try to add that to your question, and it might get some good answers! $\endgroup$
    – kjetil b halvorsen ♦
    Commented Nov 2, 2016 at 9:21
  • $\begingroup$ @kjetilbhalvorsen Ahh.. I get it. Because I don't know anything about statistics... I asked a question vaguely. Above data is the outputs of the program A and B. Because the program is probabilistic, it always results in different output. I assume that I do not know whether program A and B is same, but I want to guess program A is same one as B, from the results. If the results are too different, I conclude program A is different from program B. Or, the results are quite similar(though it has different value) I conclude program A is same as B. $\endgroup$
    – cointreau
    Commented Nov 2, 2016 at 9:36
  • $\begingroup$ @kjetilbhalvorsen to answer my question, I thought I should see the form of distributions in two sets. Is it still not enough...?? $\endgroup$
    – cointreau
    Commented Nov 2, 2016 at 9:37
  • $\begingroup$ Good. New information should be added as an edit to the original post, please do so (there is an "edit" tab below the post, use that. And also, the numbers in the output from the program, what do these represent? counts? money? ... ? state that too! $\endgroup$
    – kjetil b halvorsen ♦
    Commented Nov 2, 2016 at 9:38
  • $\begingroup$ Well, yes, we might make some plots or calculate some numbers, see some difference, bt to see if that difference is IMPORTANT or not, we MUST know a context! The importance derives from that context, not from the numbers. $\endgroup$
    – kjetil b halvorsen ♦
    Commented Nov 2, 2016 at 9:40

1 Answer 1

Reset to default
0
$\begingroup$

I would start with the interoccular trauma test, in your case:

enter image description here

set A seems a bit more variable then set B, but the main thing is the huge outlier in set A.

Improve this answer
$\endgroup$
3
  • $\begingroup$ Thank you for your kind answer, but I'd like to know how different set A is from set B, not to compare variability between two sets. $\endgroup$
    – cointreau
    Commented Nov 2, 2016 at 9:20
  • 1
    $\begingroup$ Set A and set B differ with respect to their variability. If you want to know how different they are you have to define the dimension in which you want to measure it. Since you did not define it, I used the graph to find the most telling dimension. $\endgroup$
    – Maarten Buis
    Commented Nov 2, 2016 at 9:23
  • $\begingroup$ I edited my question... I want to know the difference of distribution in values between two sets. $\endgroup$
    – cointreau
    Commented Nov 2, 2016 at 9:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.