How to create shaded error bands using Excel for Mac
Вставка
- Опубліковано 19 сер 2024
- This tutorial shows you how to create shaded error bands in Excel. Shaded error bands are useful when you want to convey the error associated with points in a line graph in an elegant and clean way.
For a full, written tutorial, visit CEREBLOG at the link below:
nmarinsek.com/t...
Thank you for this, very helpful! As a note to folks who are using 2013 Excel: If you have a graph with multiple lines representing the mean and overlapping 95% confidence intervals you will need to change the order in which they appear on the graph. To alter this, change your upper and lower bounds to an Area Chart (as indicated in video). Next you can go into the "Select Data" function and move the series up or down. The higher it is on the list, the farther back it will be displayed on the graph. Hope this helps, took me a bit to figure it out.
Thank you Erin, this was driving me nuts!
Thank you thank you thank you. I was tearing my hair out trying to work out why this all went west when I reversed the axis values. I knew I had to reset the order but didn't know how. Thank you again Erin!
Erin, this is an extremely important tip! Thanks for sharing!
What if one of the lines is higher at some points and lower at other points?
You're my hero!!!!
Wherever you are in life thank you for doing this video!
Thank you very much. Appreciate it. Note that also on the newer Excel versions on the mac the way layers overlap is not okay by itself. Changing the order has to be done in the graph setup instead of the "arrange" submenu. I find it useful to name each row differently to really distinct std-dev, UL and LL for each of the topics (your courses). This makes sorting so much easier than ending up with several data lines all named LL, UL or STDEV
Very nice. Unfortunately it doesn't work if your lower bounds cross at any point.
Exctly what i was looking for. Thanks. I dont know if anyone feels the same, but your voice was so sweet! Lol.
You can add error bars to a line chart, making them Y only bars and remove the cap on them. The problem is that there are only as many error bars as data points, so to do it properly you would need to interpolate the spaces in between your data points to sufficiently fill in the error bars to make it look continuous (or make the bars fat enough to give that appearance). I actually did have continuous data (EEG, sampled at 1000Hz, so a total of about 1200 data points) so I didn't have to do that. Just thought I'd mention it. This also works to allow overlapping intervals, you just make the error bar whatever percentage works for you.
Thank you for the brilliant tutorial.
Awesome and very useful. Thanks a lot!
Exactly what I was looking for and thought I would have to program in R - thank you!!!
A great tutorial. However, I had six line graphs for which i needed error bands. It worked for about 2-3 but after that it became impossible to see anything. I finally left the old style error bars in. Is there anyway around this for multiple line graphs on one chart?
For those wondering how to do this when you have errors bans that change sign...create a 3 column (or row) using the IF statement and then shade what is not needed as white.
can you explain this further?
It was useful but does not work for the graph with both negative and positive values.
I just love you and I also love your voice
Impressive one! Thanks for the tutorial
This doesn't work for negative numbers. Does anyone know how to do this for a combination of negative and positive numbers.
Thanks!
that trick was pretty impressing as well as useful: thanks for uploading
Thank you
This method does not work for scatter plots or if the data crosses y=0. Bummer. This is not really the author's fault, just pointing out that MS should add this as a feature and not require this workaround.
Hello Russel,
Do you ahve any idea how to work for data that crosses y = 0 ?
Thanks !
Just what I was looking for. Thanks a lot :)
Fantastic thanks for this! Even though this video is old, can you by any chance advice how to do when we have negative values?
It looks like if you fill the lower limit area in the same color (instead of white as instructed in the video), you can accommodate negatives
Hi, when you have multiple series, the white lower bound area of one data overlaps some of the important parts of other data, meaning my shaded SD area is not continuous with missing parts, how to i correct this?
Great video! Thanks for sharing it!
This is excellent! Thanks!
What if one of the lines is higher at some points and lower at other points?
smart!
Thank you it was very useful
No way to have gridlines with this though
Thanks. Very helpful!
This video is already so old. Is there a way to do this automatically without having to create upper and lower columns of data? Only with data and stdev columns? Auxiliary columns will really gum up my excel with unnecessary data or extra notebooks.
Doesn't work unfortunately, if you have x-y-data as input. :-(
thank you!
Thank you :)
That's amazing thanks very much for sharing. I have a question for you though. Do you know how to deal with crossing lines? It seems that when the upper line is crossing and drops below the lower one at some point the shaded area is going away, probably because the fill of the lower band of the second line is set to white. Do you have any solution for that? Thanks a lot!
Thank you! The method I describe in the tutorial won't work for intersecting lines (for the reason you describe), but a tutorial called "Fill Under or Between Series in an Excel XY Chart" on the Peltier Tech blog may help. Near the end of the tutorial is a section called "Fill Between Overlapping Regions" that explains how to create intersecting shaded regions. I haven't used the tutorial myself, but it looks like you can use their instructions to create the overlapping error bands, and then simply add your time series on top of the shaded regions to achieve the look of error bands. I hope that helps!
Stupid microsoft ... change series chart type is not available on macos.
Eek! this is actually statistically really wrong. Please don't do this!
Would you mind explaining why?