What meaningful information can I extract from this data set? Sorry in advance for this question. Probably is not meant to be here. If it is not, please point me in the right direction. This is going to be very generic and probably opinion-based. However, I don't know who else to ask. 
I am not a data scientist, or statistical trained person. I am a developer that has been tasked to show "useful information" from a particular data set.
In our shop, we process documents. We take them from point A to B. This translates to take each one of the pages of the document and process them. This process is achieved using multiple threads. So, I have access to the following information:


*

*I can easily calculate the time that a page spent going from A to B in the entire process. This could be used to give some type of average time spent for any page of any document between a particular set of dates.

*Because this process is done using threads, I have access to the same information for each of the threads. For instance, thread 1 spends an average time of 2 seconds per page while the thread 2 spends 3 seconds per page.


All of this information is somewhat related, but at the end it isn't. What do you guys think that I could do with this as an analysis point of view? Is there an index kind of variable that I could calculate dividing this numbers to create some kind of ratio and relate this to something else to see if we can make decissions based on all of this?
Thanks
 A: This is difficult to answer since you're not asking a specific analysis question. Expanding on my comment, the first thing I would do is come up with a few specific questions. I'd write them down in jargon-free, English-language sentences and think about them for a while. I'd discuss them with co-workers. Have they been clearly answered before? Are they really that interesting?
I would see if I have the data to answer those questions. It might not be obvious whether you do or not. If in doubt, ask each question on Stack Exchange where you state your question, the data you have, and any attempts you've made and the results. If you don't have the appropriate data for your question, devise an experiment and collect it. If that's not possible, move on to the next question.
Now, as far as actually answering a question... In the comments you ask "Why is my system running slow?" A quick analysis you could do is:


*

*Identify when your system is running slow. This can amount to creating a label variable L, which is 0 when the system is fine and 1 when system is slow.

*Create X0 = data(L==0), X1 = data(L==1).

*Plot histograms of X0 and X1 and visually inspect for differences. (This will depend on the data, but you could start by making a single histogram for every variable in your data.)

*Do any of the differences you identified in 4 seem like they could be causing the slowdown?
i.e. Visualize your data during slowdowns and during normal operation. Are there differences?
In general there are many ways to treat the data, but they depend on exactly what the data looks like (how many dimensions, categorical or numerical, data size, etc.). You could try including very specific information about the data you have, either in this question or in a new one.
