Link to home
Start Free TrialLog in
Avatar of trcamp
trcampFlag for United States of America

asked on

normal distribution plots of column of excel data

Hi,

I have numeric data in an excel spreadsheet that I would like to disaggregate on a categorical variable and plot the resulting sets of data as probability distributions.  

The worksheet contains about 1958 rows of data.  The worksheet will be disaggregated on the dependent variable called "radon" (an integer) based on the categorical variable "code" (a text field).  Both are highlighted in the attached spreadsheet.  This will produce several sets of rows of data (I think there are about 20+ different "code" values).  For each set of rows, each set corresponding to a different code, I need to compute the median, max, min, mean, and standard deviation for that set of rows, along with a plot of its probability distribution bell curve.  This will be repreated for each of the disaggregated codes (that is, one set of statistics and one plot per code).  Finally, all of the plots should be placed on top of one another to allow me to observe the spread of each of the datasets.  

Attach is a spreadsheet with the data.

Thanks!

TC
expertexchange-radon-vs-code.xls
Avatar of Steve
Steve
Flag of United Kingdom of Great Britain and Northern Ireland image

Have you tried Minitab? It is specifically designed to do just what you want there and has a 30 day free trial available.
Avatar of trcamp

ASKER

No, but thanks for the heads up!  I am hoping to have a permanent solution using excel.  

Question still open.
As the file is saved xls I assume Excel 2003?
As this can be done simply enough with sparklines in 2010.
Attached is a rough use of PivotTable and some charting.

I am guessing that the data is not normal, so would likely require a Box-Cox to get it into a normalised form.

However, this may get you started.
expertexchange-radon-vs-code.xls
Avatar of trcamp

ASKER

Nice quick work batman.  a couple of thoughts.

1.  the plot's y axis only goes up to 25.  It should go up to around 5000 or 8000 (the mean of the set of highest radons).  

2.  Could you include the Box-Cox for more points (the max 500)?

3.  Could you allow the x axis to show the category variable ("code")?
Avatar of trcamp

ASKER

One correction.  The X axis is the one that should have the 5000 to 8000 mean value, with spreads of 2 SDs....not the y axis.
OK, have "done some work" on the file.

1) I have added the Radons to the axis

2) I have double clicked one of the Pivot lines to extract just that "code"
    This I have performed a "basic" Box-Cox Log normal transform.
    ou should be able to do the same with other lines and then copy the formula and graphs.
    Should get you going in the right direction

3) The "code" is each series line... have added a key table, but this may be more confusing than helpful.
expertexchange-radon-vs-code.xls
Avatar of trcamp

ASKER

Looks great.  One last question.  What are the "divide by zero" cells in the pivot worksheet?  I just want to make sure that those are not being calculated incorrectly or are otherwise important in some way.

Thanks.
ASKER CERTIFIED SOLUTION
Avatar of Steve
Steve
Flag of United Kingdom of Great Britain and Northern Ireland image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of trcamp

ASKER

Great.  Thanks Batman.  Nice, efficient, quick work.