I have data for 131,653 users about how often they do a task, and how much time they take when doing the task. I simply arrange the number of times they do the task in ascending order of the time they take, and make a scatter plot. This is what I get:

enter image description here

I don't understand what to make of this pattern. I have not seen anything like this before. Can anyone help me understand what might be going on?

I can zoom into the graph to show the plot for users 10,000 to 20,000:

enter image description here

And also, this is the zoomed in plot for uers 130,000 to the end:

enter image description here

EDIT: The dataset looks something like this:

User   Time     Number of
ID     (sec)    occurrences

1       110     342
2       172     332
3       193     357
4       196     492
5       206     287
6       246     338
7       270     296
8       357     440

The time taken has been sorted in ascending order, as shown in the dataset, and the users have been marked with IDs according to the time. And then, I do the scatter plot of column 3, which is giving this bizarre pattern.

EDIT 2: Let me explain a bit about the context of the data. The task is opening of notifications and the time is the response delay (i.e., how long after receiving the notifications does the user open it). So, the response time ranges from 0 to 168 hours (i.e., 1 week, which is the duration for which the data was collected - and some notifications were never opened). So, I just took the raw data, sorted the records in increasing order of the response time, and plotted the number of times each user opened a notification. Hope the context is clearer now.

EDIT 3: Here is the graph of the number of occurrences vs the response time (note that the x-axis is in log scale, as in the linear scale all the points were cluttered near the origin):

enter image description here

  • $\begingroup$ what does it mean ''in ascending order of the time they take'' ? $\endgroup$
    – user83346
    Oct 28 '16 at 13:11
  • $\begingroup$ @fcop I have edited the question to explain the data. $\endgroup$ Oct 28 '16 at 13:31
  • $\begingroup$ Are all users doing the same identical task ? $\endgroup$
    – user83346
    Oct 28 '16 at 13:46
  • 2
    $\begingroup$ Some of what you are seeing could be due to discretization of time $\endgroup$
    – bdeonovic
    Oct 28 '16 at 13:49
  • 1
    $\begingroup$ Like others I don't understand all the description here but I suspect a side-effect of using ratios based on integers. Only certain values are possible. $\endgroup$
    – Nick Cox
    Oct 28 '16 at 13:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.