## Which statistical test is significant here? [closed]

As part of a performance evaluation of a system, I have the following piece of information:

year | total jobs submitted | failed jobs


I have the same piece of information after introducing a new update to the system as a result of which the number of failed jobs decreased. The data looks again like the following:

year | total jobs submitted | failed jobs


Are there any useful metrics that I can use to say something good about the update that I used on the system?

-
What you said - "the number of failed jobs decreased" - sounds pretty comprehensive. I think your questions is too vague to have a more detailed answer, which is why nobody has given one. (Also any answer that is given will likely be at too elementary a level for this site.) If you had several years worth of data, or were able to make assumptions about the sources of variation in job failures, you could perform tests to see whether the decrease was within the natural variation of the data or whether it could confidently be attributed to your change. – Tom Smith Apr 19 2010 at 9:23
Thank you for your time. Actually I haven't been in touch with statistics for a long time so that explains the reason behind my bad question. Could you kindly elaborate on the tests that you are mentioning? – Legend Apr 19 2010 at 15:34
As stated, this question is too vague for me to consider it mathematical, so I'm voting to close. – Steve Huntsman Apr 27 2010 at 13:55
MO is intended for questions of interest to practicing mathematicians. This question is thus not appropriate and I'm voting to close. I wish I could send you to a webpage that was more appropriate, but as far as I know there isn't one. – Noah Snyder Apr 27 2010 at 14:43
Am casting the final vote to close (see previous two comments); but if people have specific and relevant references that they think could be given as useful answers, then they should vote to re-open, or contact a moderator, or both. – Yemon Choi Apr 27 2010 at 17:33