Recently, I had to make a bunch of panel charts. After wrangling with Excel (and a tiny bit of VBA) to create them, I wondered if we are suffering needlessly by being too loyal to Excel. I switched to R and could create these panel charts in almost no time (well, first I had to learn how to pivot the data using dplyr). Today, let me share the experience.
The scenario – Visualize employee performance ratings
Around July, most companies in NZ have their employee performance reviews (or rem reviews). In August, all this review data is, well, reviewed so managers can decide whether any moderation (curve fitting) is needed and how much hike to offer. To help with this process, you need to first understand what proportion of staff are in each rating category. If you are head of a large company, then naturally, you want to know how this spread looks like at overall level as well as individual business group, branch levels, ie at various hierarchies.
The solution – Panel Charts
One quick and easy way to do this is by making lots of charts.
Enter Panel Charts.
Let’s say you have 80 branches, spread across 10 business groups. Each branch has different number of people. To understand if the ratings are normally spread in a branch, you can make a column chart showing number of people in each rating category. Then to see how this spread looks across the company, simply create 80 charts, one per branch.
Related: Introduction to Excel Panel Charts.
80 Charts – that is not fun!!!
Tell me about it. You might need a strong cup of espresso and right mood for lots of copy pasting.
Or you could use some VBA to do the dirty work for you.
In either case, you are going to waste (and hate) that afternoon.
An alternative – use Sparklines
If you are not in the mood for manic copy pasting, then you could use sparklines to cheat. Once you make 80 sparklines, you can just move them around so they look like an 8×10 grid.
It is still a lot of work, but less annoying than creating 80 charts.
A better solution – use R
If you are familiar with programming (VBA/C or something else) you can quickly learn the basics of R and use simple code to convert raw data to elegant panel chart in almost no time.
OK, show me how.
Watch below video. I explain both Excel sparkline and R approaches. As the process is elaborate (especially the Excel one), I am using video format. See it below or on Chandoo.org channel.
Download the resources – Excel file, R code and CSV
Use below links to download files:
- R code panel-gen.r
- Excel workbook with sparkline panel charts – panel-charts
- Raw data in CSV format – rem-data
Have you used R yet?
R is fun, interesting and challenging. If you have some free time, pick it up. It is free and offers heaps of possibilities for statistical analysis, data science, modeling, visualization, data cleanup and exploration. If you are looking for a book, I suggest R for Data Science by Hadley Wickham and Garret Grolemund.
If you are doing some R programming, please share your experiences and thoughts in comments.
18 Responses to “Best Charts to Compare Actual Values with Targets – What is your take?”
Great post. I can't vote, though, because the answer I want to put down is "it depends". As with all visualisations, you've got to take into account your audience, your purpose, technical skills, where it will be viewed, etc.
I'm with Andy: It depends. Some I would use, some I might use, some I won't touch with a barge pole.
Naturally I have comments 🙂
The dial gauge, though familiar, is less easy to read than a linear type of chart (thermometer or bullet). It's really no better than the traffic lights, because all it can really tell you is which category the point falls in: red, yellow, or green.
By the same token, pie charts are so familiar, people don't know they can't read them. Remember how long it takes kids to learn to read an analog clock?
Bullet charts don't show trends.
With any of the charts that have a filled component and a marker or ine component, it makes more sense to use the filled component (area/ column) for target, and the lines or markers for actual.
[...] Best Charts to Compare Actual values with Targets (or Budgets … [...]
I voted for #6 even though I agree with the other comments that it depends.
The majority of the votes are for the #2, thermometer chart. I still have yet to understand what happens when you are above plan/goal, which was brought up in yesterday's post.
Also, I agree with Jon in that it would be better to flip the series and make the filled part the target or goal and the line or marker the actual.
I am also a fan of using text when appropriate if the data is among other metrics in a type of dashboard. Calling it out by saying actual and % achievement is a good option.
Another "it depends" vote. Are you just looking at one or are you comparing a number of targets with actuals? You didn't include a text box. The problem with sentences is that they can get lost in a page of gray text. A text box can call attention to the numbers and line them up effectively.
I'm with Jon: "Some I would use, some I might use, some I won’t touch with a barge pole" and I'm surprised that some of your readers voted for the last group.
Jon says:
With any of the charts that have a filled component and a marker or line component, it makes more sense to use the filled component (area/ column) for target, and the lines or markers for actual.
Why does this make more sense? I like 6 the way it is, although I would use a heavy dash for the plan/target marker.
"It depends" is also my take. What I usually try to drill into my clients dashboard design is the fu ndamental difference between spot results (am I on target for this month) and long term trends.. I always try to create 3 different set of graphs to represent real perormance:
- spot results vs objectives
- cumulative results vs objectives
- long-term trend (moving average) mostly) to see where we're going
[...] Best Charts to Compare Actual Values with Targets – What is your take? (tags: excel charts) [...]
[...] Related: Charting Principles, How to compare actual values with budgets [...]
[...] Excel Charting Alternatives to compare values [...]
Jon says:
With any of the charts that have a filled component and a marker or line component, it makes more sense to use the filled component (area/ column) for target, and the lines or markers for actual.
Why does this make more sense? I like 6 the way it is, although I would use a heavy dash for the plan/target marker.
I totally agree, Bob. I would normally favour a line for the target and a column for the actual, you can see quite easily then which columns break through the line, then.
[...] best charts to compare actual values with targets — den Status mal anders zeigen, z. B. als Tacho [...]
Thermometer charts: "Not appropriate when actual values exceed targets" - this is easily solved by making the "mercury" portion a different color from the border, then you can clearly see where the expected range ends and the actual values keep going.
People seem to knock gauges quite a bit in dashboarding, but trying to show comparison of realtime data between operating sites and targets for each site can easily be done with a bank of gauges that have the optimal operating points at 12 o'clock.
The human eye is great at pattern stripping, and any deviation of a gauge from the expected 12 position will quickly register with an operator and attract his attention. Using a colour background, or meter edge, will also indicate the sensitivity of a particular site.
[…] Best charts to compare actual with target values […]
[…] Best charts to compare actual with target values […]
[…] work laptop I have a favorites folder just dedicated to Excel charts. Its got things like “Best Charts to Compare Actuals vs Targets” and “Best charts to show progress“. I love me some charts […]
I am wondering how will the plotting work, for some of the targets which may have been achieved before time. E.g. for the month of Jul the target was 226 and the actual was 219. So the chart will show a deficit in meeting the target by 7 points but what if this 7 may have been completed earlier in month of June. So ideally it not a deficit.