I have a data set with 32 variables. I take the summary of these variables. I found that some of these variables have a Mean or Median equal to zero. Is that Ok? Can I still work with this data set? Why they are equal to zero?

My data is Ionosphere (I worked with the variables 3:34). The data is available online from KEEl. The description of the data is:

Ionosphere data set

1: Description.

This data set is a modified version of the Ionosphere database present in the UCI repository. The second attribute has been removed due it only a single value (0) was present for it.

The task is to determine if a given signal is Good (g) or Bad (b).

2: Type.            Classification  
3: Origin.          Real world
4: Instances.       351
5: Features.        33
6: Classes.         2   
7: Missing values.  No

8: Header.

@relation ionosphere
@attribute Pulse1 integer [0, 1]
@attribute Pulse3 real [-1.0, 1.0]
@attribute Pulse4 real [-1.0, 1.0]
@attribute Pulse5 real [-1.0, 1.0]
@attribute Pulse6 real [-1.0, 1.0]
@attribute Pulse7 real [-1.0, 1.0]
@attribute Pulse8 real [-1.0, 1.0]
@attribute Pulse9 real [-1.0, 1.0]
@attribute Pulse10 real [-1.0, 1.0]
@attribute Pulse11 real [-1.0, 1.0]
@attribute Pulse12 real [-1.0, 1.0]
@attribute Pulse13 real [-1.0, 1.0]
@attribute Pulse14 real [-1.0, 1.0]
@attribute Pulse15 real [-1.0, 1.0]
@attribute Pulse16 real [-1.0, 1.0]
@attribute Pulse17 real [-1.0, 1.0]
@attribute Pulse18 real [-1.0, 1.0]
@attribute Pulse19 real [-1.0, 1.0]
@attribute Pulse20 real [-1.0, 1.0]
@attribute Pulse21 real [-1.0, 1.0]
@attribute Pulse22 real [-1.0, 1.0]
@attribute Pulse23 real [-1.0, 1.0]
@attribute Pulse24 real [-1.0, 1.0]
@attribute Pulse25 real [-1.0, 1.0]
@attribute Pulse26 real [-1.0, 1.0]
@attribute Pulse27 real [-1.0, 1.0]
@attribute Pulse28 real [-1.0, 1.0]
@attribute Pulse29 real [-1.0, 1.0]
@attribute Pulse30 real [-1.0, 1.0]
@attribute Pulse31 real [-1.0, 1.0]
@attribute Pulse32 real [-1.0, 1.0]
@attribute Pulse33 real [-1.0, 1.0]
@attribute Pulse34 real [-1.0, 1.0]
@attribute Class {g, b}
@inputs Pulse1, Pulse3, Pulse4, Pulse5, Pulse6,
Pulse7, Pulse8, Pulse9, Pulse10, Pulse11, Pulse12, Pulse13, Pulse14, 
Pulse15, Pulse16, Pulse17, Pulse18, Pulse19, Pulse20, Pulse21,
Pulse22, Pulse23, Pulse24, Pulse25, Pulse26, Pulse27, Pulse28, 
Pulse29, Pulse30, Pulse31, Pulse32, Pulse33, Pulse34

@outputs Class

This is an example of one variable (20 observations out of 351).

 1  0.605 
 2 -0.517 
 3  0.546 
 4 -1     
 5  0.344 
 6 -0.0457
 7  0.746 
 8  0     
 9  0.904 
10 -0.337 
11  0.907 
12  1     
13  0.937 
14  1     
15  1     
16 -1     
17  0.942 
18  1     
19  0.786 
20  0    

closed as unclear what you're asking by Stephan Kolassa, mdewey, user158565, Michael Chernick, whuber Jun 25 at 15:31

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 5
    $\begingroup$ You have provided a very little information about your data set and it is very difficult to say if the mean and median of the variables can be zero or not unless we know the variables, still I don't see why we cannot have mean and median zero. We can have mean and median zero. $\endgroup$ – Prasad Dalvi Jun 25 at 7:34
  • $\begingroup$ @PrasadDalvi Thank you for your comment and sorry for my little information. I will provide much more details about my data. $\endgroup$ – Mary Jun 25 at 7:42
  • $\begingroup$ Much better to give an example. Ionosphere and KEEI won't mean much to most readers without their doing a web search. $\endgroup$ – Nick Cox Jun 25 at 7:47
  • $\begingroup$ @NickCox Thank you for your comment. I have added a description of the data with their domain. $\endgroup$ – Mary Jun 25 at 7:56
  • 1
    $\begingroup$ Does the interval notation $[-1, 1]$ and $[0, 1]$ used by your unnamed software have the standard mathematical meaning? $\endgroup$ – Nick Cox Jun 25 at 7:59

There is no rule that the mean and median cannot be zero.

This can be expanded a little:

  1. A variable that is always zero will necessarily have mean and median zero.

  2. A variable that can in principle be only zero or positive can only have mean zero if all values in practice are zero. On the other hand, such a variable can and will have median zero if more than half of the values are zero. Such variables are very common and include counts and indicators (commonly, variables that are 1 or 0).

  3. Point 2 applies also to variables that can in principle be negative or zero, although such variables seem less common.

  4. A variable that can be negative, zero, or positive can also have mean and median zero.

  5. It's even possible to have mean and median zero for a variable that cannot be zero, as contemplation of $-1, -1, 1, 1$ will make clear.

Being puzzled by this implies that looking more closely at your data (graphs of distributions, other summary statistics) is needed.

  • $\begingroup$ Thank you so much for your help. Is that OK for 1st Quantile as well? $\endgroup$ – Mary Jun 25 at 7:48
  • 1
    $\begingroup$ I guess you mean first quartile. But all the arguments used for median apply also with modification to any other quantile. $\endgroup$ – Nick Cox Jun 25 at 7:50

If you have a random variable following a standard normal distribution, then we would expect its mean to be 0 so this is perfectly possible.


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