Re: Histogram "Bins" tab options problem

["Daniel P. Dougherty" <doughe57 msu edu>]

On Wednesday, February 09, 2011 12:57:51 am Andreas Guelzow wrote:
> Consider the following (discrete) example data:  
>      2
>     10
>     10
>      5
>      9
>      2
>      5
>     10
>      8
>     10
>
> This would be really simple:
>
> 1)  run the analysis
> 2)  format the spreadsheet cells however you like
> 3)  then create your column chart.

No it turns out it isn't simple at all.  The "Custom Formatting" can't be 
easily formatted to "Number" if the infinite bounds are used in the bins 
option. It all just turns to ####### or blows out the axis of the chart (both 
bizarre behavior)

 --> So OK, I then choose the one of the last two binning options which have 
only the finite bounds.  Either of these options  loses all of my 2's and  10's 
from my data because it doesn't capture the maximum of my data!!  My histogram 
only has 80% of my data now :(  

Histogram               
                Column 1
above 2 up to 2.88888888888889                                 0.0%
above 2.88888888888889  up to 3.77777777777778  0.0%
above 3.77777777777778  up to 4.66666666666667  0.0%
above 4.66666666666667  up to 5.55555555555556  20.0%
above 5.55555555555556  up to 6.44444444444444  0.0%
above 6.44444444444444  up to 7.33333333333333  0.0%
above 7.33333333333333  up to 8.22222222222222  10.0%
above 8.22222222222222  up to 9.11111111111111  10.0%
above 9.11111111111111  up to 10                                        40.0%


So what do I need to do then?? I need to go back scan all my data (big pain if 
it's a large data set or need to use functions) for max and min then set up 
"Calculated cutoffs" with "Minimum cutoff" and "Maximum cutoff"  some  
perturbation below my min and above max so I can get 100% of my data back into 
my bins and then I go through the whole procedure again and now essentially 
needing to by-pass all the new features of Gnumeric's histogram features (like 
select "No chart" option and do my own Column chart setting up my own labels) 
to get the most basic frequency histogram from a small data set. 

The point one comes away with is that it is not easy at all in Gnumeric to 
reproduce what is default histogram behavior of other _very_ popular software.  
I think this is a problem.