1

I am working with a dataset (90000 observations) and have the following histogram plot (100 bins):

enter image description here

The distribution looks normal. I see collections of 3-bins forming some kinds of count peaks. How can I interpret these 3-bins collections? Thanks.

Sharek
  • 61
  • 1
  • 5
  • 2
    How precise is the data? I wonder if some rounding might lead to this pattern. If so, your bin size might be too small for the data. – mkt Aug 11 '17 at 15:49
  • @mkt thanks for your comment. Not sure if rounding made on the data. Maybe during collection, but not during my analysis. – Sharek Aug 11 '17 at 15:59
  • 2
    It looks to me like the peak in the center is too high for a normal distribution. – Michael R. Chernick Aug 11 '17 at 19:10

2 Answers2

-1

May be these are outliers or you can take a mean of each bean and then draw a histogram.

enter image description here

Manish Kumar
  • 159
  • 1
  • 3
-1

Are the data truly continuous or not? If they are not truly continuous you could get the actual values and see what sense you can make of them, given context (which you didn't supply).

If you want to assess normality, a quantile normal plot is a better method.

Peter Flom
  • 94,055
  • 35
  • 143
  • 276