Pie charts, as well as many other charts, can be built to show percentages of a whole. Unfortunately, many times the pie chart percentages do not tie in total to the total of the source document percentages. Microsoft acknowledges the problem via this knowledge base article titled Pie Chart Shows Incorrect Percentage Value in Excel.

As a potential work-around, I attempted to create a source document formula that works similar to the logic described in the knowledge base article. The purpose of the formula is to force the rounded percentages in the source report to always tie to the pie chart percentages.

Refer to the screenshot below:

A source range, named "Data", covers the range C4:C10. Column D reflects the rounded source document percentages. When added together, the percentages equal 101%.

Column E contains the alternative percentages that tie to the pie chart. The formula, entered as a multicell-array into the range E4:E10, is as follows:

The formula works essentially as the knowledge base article describes. It:

• Uses the ROW function to first break any ties that may occur when the same number appears two or more times.
• Using the numbers calculated by the ROW function, it then uses the LARGE function to determain which percentage is the maximum.
• If the percentage is the maximum, it subtracts the total of all of the the percentages excluding the maximum from 100%. All other percentages are calculated and rounded as normal.

In the example, the formula recognizes the rounded value for Gamma as the maximum. Because the data does not add to 100%, the formula subtracts one from the Gamma value to force the percentages to round to 100%. Likewise, the percentages in the pie chart show the same values.

To test the formula, I used the RANDBETWEEN function and a macro to quickly test 10,000 random scenarios. The formula appeared to work well with values that round in total to 99% or 101%. However, when values round to 98% or 102%, I did find occurances where the formula returns values that differ from the pie chart. In those cases Excel does not seem to assign the difference to one single value. Rather, Excel appears to pick two of the same values and distribute the difference evenly. Given the logic as described in the knowledge base article, I'm at a loss to explain why.

Below is an example where the pie chart percentages differ from the formula:

The formula calculates Beta to be the maximum and subtracts two from it to force the total to equal 100%. However, Excel subtracts one from Beta and Delta to get to 100%.

From Microsoft - Knowledge Base Article 213644:

In Microsoft Excel, you can set the Minimum and Maximum properties of the y-axis scale to any value. The following macro sets the minimum and maximum values for the y-axis scale in a chart to the maximum and minimum values of the data used to create the chart.

It you need to use "=NA()" as a source value, the macro above fails. However, you can modify the macro to allow the array to accept only numeric values. To do so, modify the For-Next procedure as follows:

Jon Peltier provides an excellent example of how to create waterfall charts that cross the x axis on his Peltier Technical Services site. These charts are especially useful for illustrating the components of variance.

Adapting Jon's waterfall charting examples to show the components of variance in a healthcare environment (for Hospitals, Clinics, etc.) requires an extra step. Normally, a positive variance would be shown with a green bar moving up and a negative variance would be shown with a red bar moving down. In a healthcare setting there's an extra twist. A positive variance can be shown with a green bar moving up or down. Likewise, a negative variance can be shown with a red bar moving up or down. Given these additional requirements, the formulas that drive the variance bars must be modified to reflect these types of changes.

To further explain, revenue provided by commercial and contract payers allows a greater reimbursement than revenue provided by government payers. As a result, an increase in the revenue provided by a commerical and contract payer would be favorable. A decrease in the revenue provided by a government payer is also favorable. Conversely, a decrease in revenue provided by commercial and contract payers may be viewed as unfavorable while an increase in revenue provided by government payers would also be viewed as unfavorable.

Below is an example of how a healthcare waterfall chart might look:

To allow for the changes in color if the variance is positive or negative, I've added a "Yes" or "No" color switch to the range C29:C33. Rather than the values determining if the color of the bar is positive or negative, the Yes-No switch allows the user to determine whether a change in a bar is positive or negative.

The example shows an overall decrease in volume. However, although the decrease in volume is normally viewed as negative, the reimbursement associated with the decrease may actually increase. The increase in reimbursement is a result of the positive commerical and contract variance combined with a negative government variance. To summarize, even though the volumes are decreasing the net revenue per unit of volume increases which allows for the possibility of greater profitability.

Since the example is somewhat complex, rather than explaining how each of the formulas work I've provided a link to the example file below.

I've recently updated the Formula Trace Tool utility to include a few new features. Behind the scenes, the formula auditing code has been streamlined to run more efficiently. If interested, please visit the Downloads page for more information.

Recently a project required that I perform calculations on the point values of an Excel chart using VBA. Upon researching how to do so, I found that writing the code is not as easy as I originally thought. The task becomes difficult because the point object does not have a values property.

Fortunately I found a solution that was posted by Andy Pope in response to a question posted to the Ozgrid.com forum. Below is Andy's example:

In the solution above, "x" is a single-dimension array that hold the point values of "SeriesCollection(1)". A loop is used to increment through the array and return the point values.

Using this solution as a start, I needed to expand it to cover all of the series in a chart. The example below is the result:

In PointValues1, the first loop is used to increment through each series on the chart. Like Andy's example, the statement

is used to extract the point values from the active series. Because the array needs to hold all of the point values of the chart, a simple counter is used to record the placement of each point value within the array where that value should fall.

I created the following chart in which to test the macro above:

The macro returns the following point values in the immediate window from the "arrValues" array:

Next, I modified the macro above to perform several calculations using the point values stored in the array. The modified code below returns a maximum, minimum, total, and average value for all of the data points in the chart:

The macro returns the following point values in the immediate window from the "arrValues" array:

Finally, I thought to experiment with creating a two-dimension array to store the point values and a one-dimension array to store the values from the X-Axis. In this example, a counter like those used in examples 1 and 2 is replaced with two loops. The outer loop is based on the number of series on the chart and the inner loop is based on the number of points within each series. The code below is the result:

The macro returns the following point values in the immediate window from the "arrValues" array:

The need to fill a blank range with zeros often occurs when dealing with data that is downloaded from a mainframe. Recently I've seen several examples that utilize a loop structure to fill a blank range with zeros. Rather than using loops, below is an example that utilizes the SpecialCells method to fill a range.

The SpecialCells method is much faster than using a loop structure.

Before distributing a report as an Excel file, I always add or make sure of the following (in no particular order):

• A personal preference - use Arial or Verdana font. I've found these two fonts to be the most professional as well as the easiest to read on paper, on screen, or on the web.
• A descriptive title. I usually use three lines - the first is the company name, the second is the title of the report, and the third is the time period that the report covers.
• If the report contains confidential information, add the word "Confidential" to the header.
• If there are more than four or five lines or rows of information, use line numbers or alpha references on the left side of the report to quickly identify those lines or rows.
• If there are more than four or five columns of information, use column numbers or alpha references above the column titles to quickly identify those columns.
• Format dollars as dollars and statistics as numbers. A dollar sign ($) in front of any dollar value always insures that the viewer knows the difference between dollars and numbers.
• A footer that identifies the location of the report on the network as well as other information such as who created the report, the date and time created, page numbers, etc.
• Always remember to set up the print range before distributing the report.
• If the report contains cells with formulas that might are return errors, remove the errors or use conditional formatting to mask the errors before distribution.
• If the Excel file with report contains macros, I usually remove them before distribution so that whoever receives the report does not have to respond to macro prompts when the file is opened.
• Protect the sheets against unwanted changes if needed. I added this idea but rarely do this in practice. If I really want to protect an Excel based report from someone else making changes after distribution, I send it out as an Adobe .pdf file.

Am I missing anything?

As a follow-up to Align Charts to Worksheet Grid Post 1, the following procedure assigns a name, sizes, and aligns all of the embedded charts on a single worksheet. The key to the procedure is the location (range address) of the cell at the top left side of the chart.

Given four embedded charts on a single sheet, the output looks like this:

Each chart is assigned the name of the value in the cell just above the top left cell. Each chart is aligned to the top-left cell and sized as per the height and width assignments.

With this procedure you don't have to worry about naming the charts first or building loops to place them. Just create a new chart, move it to the top-left cell location of your choice, add a name for the chart above the top-left cell, and run the macro.