The chart comes from page 34 of A Preliminary Analysis of the President’s Budget and an Update of CBO’s Budget and Economic Outlook, published in March 2009 by the Congressional Budget Office.

I’ve manually digitized the data into this tall thin tower of numbers.

Here is the data plotted in a line chart.

To capture an assumed uncertainty in the numbers, I’ve set up the following table, in which I’ve assumes a minimum and maximum for January 2009 onward, and I’ve inserted intermediate columns a, b, c, and d.

When I add these columns of data as additional lines in the chart, I get the following. The lightest gray lines are difficult to resolve, but then, they are the most uncertain. This is fine, but it doesn’t have the filled area effect.

To achieve filled regions, I will use a stacked area chart type. In an adjacent range I’ve added the following table. The dates are the same, as are the minimum and midline predictions. The rest of the data are the calculated differences between each line and the previous, which capture the height of each colored band.

Using the series from Min through Max I’ve created a stacked area chart. There are my colored bands.

I changed the bands to appropriate shades of gray, and the minimum series is not filled at all. You can use any color scheme you like, and as many different colors and data sets as you need. I used shades of gray because these were already plentiful in my custom color palette.

To get the solid line for the midline predictions, add the midline data as a new series. Initially it’s plotted as another area stacked upon the rest.

Convert the added series to a line chart type, and format it appropriately (I’ve used dark gray).

That was pretty easy, and it’s got to be one of the shortest tutorials I’ve ever written.

Peltier Technical Services, Inc., Copyright © 2010.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

Based on Stephen Few’s earlier analysis of Marimekko Chart in A Design Problem, I followed up with The Problem with Marimekkos, in which I showed a couple of approaches to charting the same data, partly based on a multiple bar chart approach proposed by Few.

I have already written about overlapping bar charts in Marimekko Replacement – Overlapping Bars (Easy) and Marimekko Replacement – Overlapping Bars (Hard). This approach is compact but doesn’t seem as clear as a four-bar-chart approach.

I combined the four bar charts into a 2×2 panel bar chart alternative to a Marimekko chart, but I gave no hints about how to construct such a chart. I’ve already described How to Build a 2×2 Panel Chart, bit this chart is more complicated than the one in that tutorial.

This tutorial applies the same approach as the “hard” overlapping bars technique. In retrospect, there isn’t much difference in effort required between the “easy” and “hard”techniques, but the “hard” way is more flexible, allowing unequal numbres of rows and columns in the original data table.

As in the previous Marimekko-related examples, the data consists of this compact table.

The data layout is the same as in the overlapping bar chart. It is a sparse array, with the data staggered to produce the arrangement of bars in the chart.

Construction begins in the same manner as in the overlapping bar chart. The data in the table above is used to create a clustered column chart. The gap width is set to zero and the overlap is set to 100.

Here is a view of the custom color palette I used to format the bars in this chart. Bar_Chart_Graded_Colors.xls is a small downloadable Excel 2003 workbook that uses this palette.

The Company Total and Segment Total series are moved to the secondary axis. Only the tallest of the individual primary axis series peek over the tops of the Total series.

The two Y axis scales are adjusted to offset the primary and secondary bars. The secondary axis is given a minimum of -$3 million and a maximum of $3 million, so the bars occupy the upper half of the chart. The primary axis is given a maximum of $3.5 million, more than twice the maximum value on the primary axis, so the primary series occupy the bottom half of the chart. These values can be adjusted to make one row of panels taller than the other.

When the Total series were assigned to the secondary axis, Excel only provided a secondary Y axis. A secondary X axis now must be added. Since its default position is at the secondary Y maximum (the top of the chart), the bars seem to hang from the top to their values.

The secondary Y axis is formatted to make the secondary X axis cross at zero, which is in the middle of the chart.

All of the default axis labels will be replaced with customized labels, so these default labels are removed.

A number of helper series will be added to the chart. These will be XY series, and the X values are defined as the positions of the points along the category axis. This particular chart has 33 category slots as shown below. The XY sereis can be located anywhere along the axis, not just on the integral positions indicated. For example, the left Y axis is located at X=0.5, the gap in the middle is centered on X=17.5, and the right Y axis is located at X=33.5.

Data for the vertical divider is shown in teal text and for the category labels in purple text in the grid below.

The Divider data is copied and pasted into the chart as a new series. It turns into another column series on the secondary axis.

When the series is changed to an XY type and moved to the primary axis, it fits between the groups of bars, and reaches from the top to bottom of the chart.

The first two columns of the Axis data are copied, then pasted in the chart as another new series. Since the previous series was converted to an XY series and moved to the primary axis, this new series is added as an XY series on the primary axis.

Using Rob Bovey’s Chart Labeler, a free and very well-designed Excel add-in, add the labels to the Axis series. The Chart labeler can be downloaded from Rob’s Applications Professionals web site

Here’s the chart after a bit of formatting.

Data for four sets of Y axis labels are shown below. The top and bottom rows of panels each have labels for values and for percentages of the total.

As before, each block of data is pasted in the chart as a new series and assigned to the appropriate axis. The data labels on the two Values series simply use the built-in Values option for data labels. The Percentages series are labeled using the Chart Labeler. The sereies and labels in the chart below are formatted with the same colors as the text in the table above.

Here’s the chart after the labels and axis series are formatted appropriately.

The method described in Double Legend in a Single Chart can be used to add legends to the bottom panels.

By adjusting the Y axis scales, the relative heights of the two panels can be adjusted. For example, in the following chart, the maximum of the primary Y axis was changed to $2.3 million, and the minimum of the secondary Y axis was changed to -$7.5 million. The Y value of the uppermost point of the Divider series also had to be changed to $2.3 million. Everything else in the chart updated automatically, although the Y values for the legend markers were adjusted, because stretching the panel spread them too far apart.

Peltier Technical Services, Inc., Copyright © 2010.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

Like most advanced charts, you need to rearrange the data in order to make it work. Here is the original asymmetric data and a Marimekko chart made from it. The overlapping bar chart is one graphical technique designed to minimize problems with Marimekko charts.

The first step is to arrange the data into the following sparse array. The gray rows are blank; I’ve only colored them to clarify the data structure. Notice that there are two blank rows between the column (Company) and row (Segment) data. This provides wider spacing between these sections of the finished chart.

Create a column chart from this data, with series in columns. To make sure Excel parses the data properly (one row for series names, one column for category labels), type a letter (“x”) into cell A20, so there’s only one blank cell in the top left of the range. Then create the chart, and finally delete the character you types in A20. Set the Overlap option to 100 and the Gap Width to zero. In the first chart I’ve set the Gap Width to 10, so it’s obvious that the large fat bars are really comprised of several narrow bars next to each other.

Here’s a glimpse of the custom palette I used in Excel 2003 to color the bars. Bar_Chart_Graded_Colors.xls is a small downloadable Excel 2003 workbook that uses this palette.

Here’s that chart with a Gap Width of zero. Look at the perfect alignment, in contrast to the slight mismatch in the “easy” version of this chart.

The data is plotted nicely, but those category labels aren’t so nice. Set up the following data for category labels. We’ll make a dummy XY series, add labels, and hide the data points.

The X values in this table are chosen to align the markers with the clusters of bars. The numbers start with 1at the first category position (which has no bars: see below), and increment with each slot along the axis. To center a label below the first cluster of three blue bars, the point must have an X value of 3. To center a label below the first cluster of four red bars, the point must be between categories 20 and 21, so it must have an X value of 20.5. The Y values place the markers along the bottom of the chart.

Copy the first two columns of this data (not the actual labels), select the chart, and use Paste Special to add the data as a new series.

The series initially does not appear. Select one of the visible narrow series and pressing the up arrow key until the new seeries is selected. You’ll probably be able to see a line of selection handles along the left portion of the bottom axis.

Use Chart menu > Chart Type or Chart Tools > Design ribbon tab > Change Chart Type to change this series to an XY type series. It now appears as a set of markers along the bottom of the chart, and Excel has conveniently added secondary axes for the new series.

Well, in this case, not so conveniently. Format the new XY series to associate it with the primary axis, and the markers now line up perfectly with our clusters of bars.

If you’re a regular reader, you already use Rob Bovey’s totally excellent Chart Labeler. If you’re not using it already, then run, don’t walk, to Rob’s Applications Professionals web site, and download the Chart Labeler. Rob calls it the XY Chart Labeler, but it works for any chart type that supports data labels. It’s free, easy to install, and easy to use.

Using the Chart Labeler, add the labels from the third column of the data range shown above to the points of the XY series. For clarity I’ve colored the labels green to match the points and to distinguish them from the default labels, which we are going to jettison in a moment.

There’s just a little cleaning up. First, hide the XY series by formatting it to show no markers and no lines. Change the labels we’ve recently added to black text. Format the category axis to use no axis tick labels. This removes the margin below the plot area, and our added labels float above the bottom edge of the chart. Resize the plot area to regain this margin and provide room for the labels.

Is the “hard” overlapping bar chart method really more difficult than the “easy” method. It’s certainly more flexible.

To construct a 2×2 panel chart that shows the same data, perhaps somewhat more clearly, see Marimekko Replacement – 2 by 2 Panel.

IIf you still want to make a Marimekko chart, you might consider the PTS Marimekko Chart Utility. Like all of my commercial utilities, it reduces the manual process of constructing a complex chart to simply selecting a range, choosing a couple of options, and admiring the result.

I’ve recently developed the PTS Merimekko Chart Utility. Like all of my commercial utilities, it reduces the manual process of constructing a complex chart to simply selecting a range, choosing a couple of options, and admiring the result.

Peltier Technical Services, Inc., Copyright © 2010.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

The approach shown here is somewhat simplified, and works only when the grid is symmetrical, with the same number of rows and columns.

Like most advanced charts, you need to rearrange the data in order to make it work. Here is the original data and a Marimekko chart made from it.

The first step is to compile the row and column subtotals into columns, with row and column headers in the first column, and the data staggered so each set of data appears in its half of the chart (in this case, Company to the left and Segment to the right).

Make a clustered column chart with these three columns of data.

Set the Gap Width of these series to 25, so the gaps between clusters are 25% as wide as the bars. This 25% comes from the final chart, which will have four narrow columns per wide column, with a narrow-column-width gap between.

Set the Overlap to 100, so the bars of one series overlap with the blanks in the other.

Now rearrange the rest of the data into the following sparse array. The gray rows are blank; I’ve only colored them for my own benefit in my working file.

Copy this range, select the chart, and use Paste Special to add the data to the chart as new series. Because the Overlap was set to 100, these new bars overlap the existing ones. For now I’ve colored the bars in contrasting colors with a custom palette.

Here’s a glimpse of the custom Excel 2003 color palette I used to format these bars. Bar_Chart_Graded_Colors.xls is a small downloadable Excel 2003 workbook that uses this palette.

One by one, move each of the series you’ve just added to the secondary axis. Change the Gap Width of one of the added series to 0 (zero). This is as good a time as any to remove the legend.

Excel has added the secondary Y axis, but we want the secondary X axis.In Excel 2003 and earlier, go to Chart menu > Chart Options > Axes tab, uncheck the Secondary Value (Y) Axis box and check the Secondary Category (X) Axis box. The bars almost line up.

Format the secondary category axis (top of the chart) so the value axis does not cross between categories. Now the alignment is almost perfect. Tweaking the width of the chart by a couple pixels may make the alignment slightly better, or slightly worse.

Changing to a matching color scheme completes the chart, and makes the misalignment almost vanish.

Changing the gap width of the secondary axis bars to 10 minimizes the misalignment and provides a border between the small bars.

The more complicated approach allows for asymmetric data arrays (different numbers of rows and columns) and also eliminates this slight misalignment. The next tutorial, Marimekko Replacement – Overlapping Bars (Hard), will describe that method for making overlapping column charts.

To construct a 2×2 panel chart that shows the same data, perhaps somewhat more clearly, see Marimekko Replacement – 2 by 2 Panel.

IIf you still want to make a Marimekko chart, you might consider the PTS Marimekko Chart Utility. Like all of my commercial utilities, it reduces the manual process of constructing a complex chart to simply selecting a range, choosing a couple of options, and admiring the result.

Peltier Technical Services, Inc., Copyright © 2010.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

This tutorial shows how to create Marimekko Charts, including the specialized data layout needed, and the detailed combination of chart series and chart types required. This manual process takes time, can be prone to error, and becomes tedious.

I have created the Peltier Tech Marimekko Chart Utility to create such charts automatically from raw data. This utility, a standard Excel add-in, lays out data in the required layout, then constructs a chart with the right combination of chart types. This is a commercial product, tested on hundreds of machines in a wide variety of configurations, which saves time and aggravation.

Please visit the Peltier Tech Marimekko Chart Utility page or the Peltier Tech Marimekko Chart Utility Documentation page for more information.

In a recent tutorial I showed how to make Variable Width Column Charts. A stacked version of this chart type, called a Marimekko chart, is popular in business, particularly where marketing departments analyze segmentation of their markets. The Marimekko chart gets its name from the patterns found in Marimekko fabric.

Marimekko Chart

Laying Out the Data

Suppose we have four companies, Alpha through Delta, competing in four segments, A through D, of a particular market. This could be like the companies Dell, HP, and Toshiba competing in the laptop, desktop, and home server segments of the personal computer market. A summary of the data would look like this table. The sum of the segment percentages is 100%, and the market segment shares of the companies in each segment (each row) also sum to 100%.

Like most non-standard charts, we have to arrange the chart source data in a special pattern. Here is how the above data is rearranged. The first column, the X values, comprise a cumulative percentage of the segment percentages: 0, 40, 70 (40+30), 90 (40+30+20), and 100 (40+30+20+10). Each of these values is listed three times. In between each of these values, I’ve inserted another row with the average of the values above and below, with orange text. These orange values correspond to dummy data which will provide labels later in this protocol.

Note that the X values look like percentages, but are not. A date-scale axis, which we will use to spread out the widths of the area chart data, cannot render fractional values. These X values are whole numbers, and I’ve used a number format of 0"%" with the percentage symbol in quotation marks to place it as generic text after the whole number.

The Y values for each series and each segment are repeated three times, at the beginning, middle, and ending X values for this segment. For example, Alpha has 60% for Seg A in rows 3 through 5, next to X values of 0%, 20%, and 40%. The Y values return to zero at the end of each segment. Alpha shows 0% in rows 2, 6, 10, etc.

The column labeled “dummy” holds Y values for a dummy series that will provide the labels in the column marked “Labels”. I’ve colored this text orange to indicate labeling helper data, corresponding with the inserted orange rows above.

Creating the Chart

Select the data in columns H through M and create a stacked area chart.

Convert series LabelsY into a line chart series.

If you clear the zero values in the LabelY data, column M, the points and connecting lines will not clutter up the chart.

Clear the legend to add more room for the chart.

Now the X axis needs to be adjusted. All the X values in column H are treated as text labels, equally spaced without regard to their numerical value. In Classic Excel (2003 and earlier) go to the Chart menu > Chart Options > Axes tab.

The Primary Category axis is listed as Automatic. Change this setting to Time-Scale.

The result is that all Y values for equal X values are vertically aligned, as if in an XY chart. Of course, Excel has helpfully converted the values into dates, spaced one week apart.

Format the X axis scale so it runs from 0 to 100 in steps of 20 (not percent but whole numbers). Apply a number format of 0″%”, which places the percent sign in quotes so Excel adds it as text to the end of the whole number rather than converting the value to a percentage.

Add Labels to the Marimekko Chart

To add segment labels, I use Rob Bovey’s free Chart Labeler utility. This utility adds a menu item called “XY Chart Labels” to the Tools menu. It works on any chart type that accommodates chart labels, not just XY charts.

Using the utility, assign the labels in the Labels column to the LabelY series. The labels here have been positioned above the points.

Format the Labels series so it uses no markers or lines, and these points will no longer appear.

To add labels to the blocks in the chart, you need points which will be located in the center of each block. The X values are the same as above, but the Y values for each series (the names are the same as above, but they are distinct series) are calculated to place a point at the mid-height of each block.

Copy this data, select the chart, and use Paste Special to add the data as new series. Since the last series in the chart, LabelsY, had been converted to a line type, these are added as lines. Notice that the points are located exactly where needed, although there are also points along the category axis and lines everywhere.

Clear the cells with zero values. . .

. . . and only the desired points will remain in the chart.

Format each series so it uses no lines or markers, and so it also displays data labels using the series name. Then format the labels so they are centered on the points (the default is to the left of the points).

This chart has way too many labels. We could get away with one label per series. I’ve removed all Y values except for the ones that put points in the largest block for each series.

This is a much cleaner chart.

Naturally you could tailor the labels to show a dollar value or a percentage, or any other information, but you would need not only a table containing the X and Y values as shown in the table above the last, but also a table containing the values to be used for each label. You would use Rob Bovey’s Chart Labeler to apply the labels, as described for the segment labels along the top of the chart.

Peltier Technical Services, Inc., Copyright © 2010.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

Several months ago, in Column Chart to Replace Multiple Pie Charts I showed how a column or bar chart could display a table of data more effectively than four pie charts could. Another alternative is to build a panel chart, a shown above.

Let’s review the sequence. The data shows how four business measures (Hours, Cost, Profit, and Sales) are apportioned among four categories of employees (Engineer 1 and 2, and Marketing 1 and 2).

Here are the pie charts. Very pretty, colorful and symmetrical and all. But it’s not really easy to compare the actual wedge sizes within and between the pies.

A clustered column chart plots all data on the same baseline, and the value is proportional to the one-dimensional size of the bars, not to a two-dimensional arc length.

The column chart could be improved by removing the legend and labeling chart series directly. In order to prevent labels from overlapping other labels and other series, it may be necessary to rotate the labels into harder to read orientations. However, we can rotate the whole chart, and retain legible labels in a clustered bar chart.

Finally, for some situations, a panel chart may be an improvement over any of the preceding alternatives. This article shows how to construct a 2-by-2 panel chart. The procedures are easily scalable to any practical R-by-C arrangement.

Panel Chart Data

The original data is shown in the heavy border in B2:F6. This data needs to be rearranged and augmented before we can create the chart.

The data is color coded as follows:

• Green – used for series names in the chart
• Purple – used for series category labels or X data
• Blue – used for series Y data
• Yellow – used for custom data labels on data points
• Orange – user inputs for on-the-fly customizations

Note: I use Rob Bovey’s Chart Labeler to apply custom labels from a worksheet range to a chart series. This is a free Excel add-in which interfaces so well with Excel’s interface that it seems built in. Rob’s utility actually links each data label to a cell, so if the cell’s contents change, the label also changes. The utility also applies the font formatting of the cell to the label, so these ranges in the data below have different formatting from the rest of the worksheet.

The data in B11:M16 will produce the bars that display the actual data, and C8:M9 contains the labels that will show the employee category in the base of each bar. To ensure that Excel correctly parses the data when you create the chart, select cell C11, type a space character, and press Return. This makes Excel think there’s a label in C11, so it starts tie category label range in C11 instead of D11.

The value in cell H3 is the maximum sized bar that will fit vertically in each panel. The largest value in the table is 45.3%, which is bumped up to 50%, then an additional margin is added for the titles in the upper end of each panel, making this height factor 0.6 or 60%.

Range B18:D23 contains the data needed for the XY series that will locate the titles in each panel. The factor of 0.85 in cell D18 means that the titles will be located at 0.85 (85%) of the way from the bottom to top of each panel. The purple values were derived as follows. The column chart has columns in slots 1, 2, 3, 4, etc. Slots 1, 6, and 11 (columns C, H, and M) have zero-height columns, and are located at panel boundaries. We want the titles centered within each panel. An XY point plotted on the category axis is positioned according to the slot numbers. The X positions for the XY points are then (1+6)/2 = 3.5 and (6+11)/2 = 8.5.

Range F18:H24 contains data to define the vertical axis and its labels. The labels in H19:H21 are selected by the user, and axis ticks and labels will be placed at the appropriate positions. The X values of 1 in F19:F24 were derived as follows. The column chart has columns in slots 1, 2, 3, 4, etc. A zero-height par (column C) is plotted in slot 1, along the left edge of the chart. Therefore we want the axis at X=1.

These ranges are not populated by hand, of course. I’m too lazy to keep changing everything when I make a minor adjustment to the chart. The following table shows which formulas to use. Select the indicated range with the first cell of the range as the active cell, type in the corresponding formula, then hold CTRL while pressing Enter.

Building the Panel Chart

The first step of the process is to select B11:M16 and create a stacked column chart (below left). Notice how the data arrangement in B11:M16 and the formulas for the “blank” series lead to the appearance in the chart.

Make the “blank” series invisible: format it so it has no border and no fill. Now the Sales and Profit bars float above the Cost and Hours bars (below right).

Notice the margins on the left and right of the plot area (left of the first E1 and right of the second M2 categories) are the same width as the gap between the first M2 and second E1 categories. We want the left and right margins to be half as wide as they are. This is easy, simply format the category axis so that the box for “Value (Y) axis crosses between categories” is unchecked (below left).

We’re not going to use the axis labels that Excel wants us to use, so let’s clean up the axes. Format each so that no axis features are visible: no line, no major or minor tick marks, and no tick labels. Delete the legend as well (below right).

The plot area now expands to fill the chart area. Select the plot area, and drag its left edge to the right to make room for the axis labels we will be adding. Drag the top and bottom edges toward each other slightly. Add panel boundaries. First format the value (Y) axis so it has fixed (not auto) values of 0 and 2 for min and max, and 1 for major unit. Format the catebory (X) axis so there are 5 categories between tick marks. Then add major horizontal and vertical gridlines, matching the line color to the border color of the plot area.

The next few steps will add the panel titles. First copy the range C19:D23, then select the chart, and use Paste Special to add the data as a new series, with series data in columns, categories in the first column (don’t replace existing categories), and series names in the first row. This produces the red-bordered stack of columns that fit poorly into the chart below at right.

Right click this new series, and choose Chart Type (or Select Chart Type) from the pop-up menu. Select an XY type. The series has converted itself to the backwards “Z” in the left hand chart below.

Excel ever so helpfully placed this XY series onto secondary axes, but we will keep it simple stupid by formatting the series so it plots on the primary axes with the stacked columns.

Cycling the chart’s secondary axes on and off have messed up the primary category axis. Select it (click on it carefully, it’s still there), and again uncheck the “Value (Y) axis crosses between categories”box (below left).

Time to fire up Rob’s chart labeler. Select the chart, then choose “XY Chart Labels” from the Tools menu (or from the Add-Ins tab in Excel 2007), and select the Add Chart Labels command. In the Labeler dialog, select the Titles series, the range B20:B23, and the Center position. The labels now appear centered over the XY markers (below right).

Hide the Titles series by formatting it to show no markers and no lines. You can adjust the vertical position of the titles by changing the value in cell D18.

The next several steps add the Y axis labels. Copy F18:G24, select the chart, and add the data as a new series using the same options as before. The new series appears along the Y axis. Since the last chart was converted to an XY series and moved to the primary axes, these are the settings that Excel uses for the new series.

Use the Chart Labeler to add the labels in H19:H24 to the newly added Axis series, in the Left position.

Format the Axis series to hide the line, and to mimic axis ticks with cross markers that match the color of the plot area border.

Finally, add the employee category labels to the bars. Use the chart labeler to apply the labels in C8:M8 to the Hours and Profit series, and the labels in C9:M9 to the Cost and Sales series. I used the Inside Base orientation.

You can download a zip file containing my worksheet with the 2-by-2 panel chart in this tutorial.

Peltier Technical Services, Inc., Copyright © 2010.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

Federal Spending on Education and Training

The first compares two years of Federal education spending data using side by side pie/donut charts, which commits two sins: using one pie/donut charts, and using more than one of these charts. Tony recommends using a table instead, while I’ve created a line chart instead. Tony’s table and my chart contain all of the data from the Business Week charts, with additional display of percentage changes.

The Original Business Week Donut Charts

Tony’s Tabulated Data

The Peltier Tech Line Chart

Tony’s table is certainly the most compact display of the data. The Business Week dual donuts are eye-catching, perhaps, but the line chart shows the trends best of all.

Percentage of Children Living in Poverty

The second chart is a simple bar chart of poverty rates by state, ranked by the rate of high school graduation in each state. Tony likes this chart, and I don’t think it’s too bad either, except that it’s divided into three sections because of its aspect ratio.

Before merely duplicating the chart above, I thought the individual graduation and poverty rates deserved some investigation. I think a heat map on a geographical state map would be interesting.

Here is my rendition of the Business Week chart, differing only by not splitting mine into three panels. It’s pretty tall, but it tabulates the data clearly. I say “tabulates” because it seems to show a trend between graduation and poverty rates, it does not allow an easy determination whether there is a correlation between these rates.

Being the geek that I am (did I say I used to be a scientist?) I decided to plot the two variables against each other. As expected, there’s a general inverse relationship. As poverty goes down, high school graduation goes up. I plotted it both ways, because there’s only a correlation, and I’m sure there are conflicting models for which variable causes the other.

I picked the latter XY chart, because I think that economic stress leads to higher drop out rates, and replaced the markers with labels showing the state abbreviations. I enlarged the chart significantly to reduce overlap of state abbreviations.

This chart is more cluttered that I would like because of the overlapping state labels, but I think it’s better to show correlations between measurable quanitites, rather than to simply list them in a table, or in a chart that is essentially like a table. What do you think?

Peltier Technical Services, Inc., Copyright © 2010.
 
Licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.
 

I need to chart three groups of two bars on a axis.
-Each group is a year, 2007, 2006, 2005 etc so we can compare data from previous years
-in each group of two there is a profit vertical bar and a combined ratio percentage bar, these bars are not related to each other (one going up has no impact on the other)
-For each of these bars, there needs to be a target indicator, a small horizontal line crossing over the bar (or above it if the target was not hit). This line does not go across the entire chart, there is a separate one for each individual bar.
-For the profit bar, if it crosses above the target bar, it should be green. otherwise red.
-For the ratio/percentage bar, if it crosses above the target bar, it should be red, otherwise green

I’ve come up with the following chart:

I don’t have the particular data, but presumably it’s something like this:

If I created a chart using A1:C4, I’d get something like the following, but I’d have to manually change the colors and somehow draw lines to represent the targets.

First we need to fix up the data, then we can build the chart. Since we want the bars to change colors automatically, we need stacked columns (see the conditional charts tutorial). Since we want clustered stacked columns, we need to stagger the data and insert appropriate blank rows (see the link to clustered stacked column charts). The data range will require one row for column headers, a blank row for the space to the left of the first cluster, two rows for the first cluster of two columns, a blank row for the space between clusters, two rows for the middle cluster of two columns, a blank row for the space between clusters, two rows for the last cluster of two columns, and a blank row for the space to the right of the last cluster. I’ve set up the data as shown below:

The “a” in some cells in columns A and B are temporary placeholders, and will be hidden later. The actual data in columns C and D will not be plotted but are used in calculations. There are formulas in columns F, G, I, and J as follows:

F3: =IF(C3>=E3,C3,0)

G3: =IF(C3<E3,C3,0)

I4: =IF(D4>=H4,D4,0)

J4: =IF(D4<H4,D4,0)

I’ve color coded the cells in the table above. The blue cells are the actual plotted Y values, the green cells are used by the chart for series names, and the purple cells are used for the chart’s category axis labels. The yellow are not used in the chart, but they help Excel parse the data into Y values, series names, and category labels.

Now we can start working on the chart. Select the cells I’ve shaded in the table. Select A1:B11, then hold Ctrl and select E1:J11, so that both areas are highlighted. Create a stacked column chart:

The first thing I’ve done is colored the series: red and green as appropriate for the Profit and Ratio series, and blue for the targets.

The target series don’t help us as columns. We’ll convert them to XY series, add X error bars, and hide the markers. First, right click the Profit Target series, select Chart Type, and select the XY type (not a Line type, because this type has no X error bars, only Y error bars).

Do the same for the Ratio Target series. If you don’t do anything after changing the type of the first series, you can simply select the second and use the F4 shortcut key, to repeat the last action.

Right click on the Profit Target series, select Format Data Series, and on the X Error Bars tab, choose both directions, Fixed Value, and enter 0.5 in the box next to fixed value.

Do the same for the Ratio Target series.

Now you need a little formatting.

• Delete the legend.
• Format the error bars so they are thicker and don’t have the end caps. Color them if desired.
• Format the target series so they have no markers as well as no lines.
• Reduce the column gap width if desired (I use 50% in this chart).
• Change “a” in the table above to ” ” (a space character). Don’t completely clear the cells, or the category labels will be messed up.

This chart isn’t too bad, and it meets the specs. But not knowing the particular usage, I think it makes more sense to put the profit values together and the ratio values together, as in the next chart. I’d also make a few other formatting changes; I might even be tempted to use area charts for the targets and line charts for the actual values, and let the reader see for himself which is good or bad without color coding the results.