In honor of the upcoming 4th of July weekend, I've declared this week to be Pivot Table week.
Afterall, what's more American than slaving away on a pivot table during a holiday.
In today’s post, I’d like to share a trick that allows you to create a frequency distribution with a pivot table.
If you’ve created a frequency distribution with the FREQUENCY function, you’ll know it can turn into a hot mess. The fact that it’s an array formula doesn’t help matters. Then there's that Histogram P.O.S. you find in the Analysis Tool Pack. That doesn't make life much better. Each time you want to change your Bins, you’ve got to run through the whole 25-step process again.
No. If I have a choice, I like to use a pivot table. Here’s a quick example how.
First, you need to create a pivot table where the data values are plotted in the Row area (not the Data area).
Notice that the SalesAmount field is placed in the Row Labels area.
Next, right-click on any value in the Row area and select Group.
In the Group dialog box, set the start and end values, then set the intervals. This will essentially create your frequency distribution.
Once you click the OK button, you can leverage the result to create a distribution view of your data.
In this example, I've added the number of Transactions to the data area to get a frequency distribution of transactions by dollar amount.
The obvious benefit to this technique is you can interactively filter the data based on other dimensions like Region and Market.
Also, unlike the Analysis Tool Pack Histogram shwag, you can quickly adjust your frequency intervals by simply right-clicking on any number in the Row area and selecting Group.
good skill
you good skill too
In honor of your upcoming 4th of July weekend, I'd like to propose a bacon recipe for Saturday.
Take 10 rashers of bacon. Add a few token bits of bread, some shavings of onion, and surround with a large Turkey.
Fire up the oven, and then Excel. Enter the Turkey into the oven. Enter the Turkey's weight into cell A1 of Excel. (DO NOT mix up these steps. And don't twitter about it if you do…we want to read about it HERE, not on Debra Dalgleish's blog.)
Enter the following cooking time formula into cell $A$2 (of Excel, not of your oven, Excel is probably not installed in your oven)
=$A$1/20*4.5
Note that this is an array formula. Or rather, it's an array formula if (and only if) you're sober enough to hold down Control + Shift + Enter after you've typed it into the cell. If you're too drunk to hold down any 3 keys – let alone the 3 keys specified above, then just go ahead and press Enter. But then don't blame me if you overtype your formula later.
Now, go ahead and set your oven's buzzer for the time specified in Cell $A$2 … assuming the format of Cell $A$2 remotely resembles a time of some sort. If it's a percent, then your goose will probably be 100% cooked.
Serve with lots of beer and low expectations.
I cannot get this to work in 2007. When I select Group I get "Cannot group that section". Thoughts?
Sounds like you're trying to group dates that are not formatted as dates. Or you have a calculated item in your pivot table already. I don't think you can group in a pivot table with calculated items.
No dates, no calculated items….no luck. Thanks anyway, I just suffer through the 25-step process.
First off, great tip. Thanks!
Second, unless there's a convention I missed back in my Stats classes, don't you’re bins need to be discrete and not overlap?
This question lead me to a curious behavior in Excel. When I tried your trick I started with a simple dataset of whole numbers from 1 to 20 and created the pivot table with a group interval of 5 which gave me four discrete bins, i.e. 1-5, 6-10, etc. All is good.
Next, I changed the first value in my dataset to 0.1 which changed the bins to 0.1-5.1, 5.1-10.1, 10.1-15.1, etc. Now my bins overlap…why and why now?
Then, with bin #1 = 0.1-5.1 and bin #2 = 5.1-10.1 I changed the fifth value in the dataset from 5 to 5.1 tentatively expecting it to be included in bin #1, however Excel included it in bin #2. Is this expected and by specific design?
Thanks
Jared,
I got the exact same problem as you when using 2007 but fine with 2003. While in 2007, it seems the row data are left justified and being treated as text even though the raw data is formatted as Accounting. With the cursor on a row data and right click. The Field Setting option does not let you change the format of the numbers to Accounting, etc.
However, the order of the row data suggests Excel knows they are numbers (1000 is after 300).
Help!?!?
Very nice trick – came at the right time, to, as I needed to generate the distribution of a few sets of data, thank you!
And I got it to work with Excel 2007 with no issue, on a field which was calculated, and formatted as %. Jared, General Ledger, what are your data?
I've often wished for an option to show "empty" groups (in your example, if there was no data for the 200-300 group, the next row after 100-200 would be 300-400). Such an option would be useful for some types of frequency distributions e.g graphical stem and leaf plots.
Very helpful – exactly what I needed – thank you.
Excellent tip. Too bad I can't set the interval manually with time data either.
wonderful tip…just does what I want
thanks very much
A year later and still nice…
Thanks for posting this