Bug 135542

Version:           1.5 (using KDE KDE 3.5.4)
Installed from:    Ubuntu Packages
Compiler:          gcc 4.1 
OS:                Linux

Currently, there is no way to add Y error bars to data points in series. For almost all scientific purposes, this is an essential aspect of all graphing 
programs, as a data point without an indication of error is pretty useless.

What I would like to see is the ability to assign arbitrary +/- Y error bars, using either data from the spreadsheet or manually entered data (or, say, constants such as "5%"), and be able to set values for +/- independently, if required (which isn't often, but if you need it at some point it's essential); being able to choose whether to apply a single value to all points for the + &/or - Y error bar, or on a series or individual basis, is also important - OpenOffice already lets you draw Y error bars, but there's only one value per series and that needs manually typing in - not useful :(

For illustration, the most common use for this in the scientific world is for displaying the 95% SE of the mean for varies data series, on line graphs - i.e. each data point is an average value, and the Y error bars are the length dictated by 1.96*(STDEV(data-samples)/SQRT(number-of-samples)) (for 95% SE, anyway... 99% replaces 1.96 with another number, which I forget temporarily). 

It would be great to have some sort of mode to draw this kind of graph without needing to manually calculate the SE beforehand (i.e. when creating the graph, you'd select all the different data points and kchart would separate them into series for you, calculate the average values for each point on the X axis and plot them, then use the data to build the Y error bars for you); it would put kspread and kchart one step ahead of MS Office, OpenOffice and gnumeric. However, "just" including the ability to have Y error bars, as described above, would give kspread+kchart parity in this area, at least.

I spoke with ingwa on #koffice, and s/he was of the opinion that "something" like this would most likely appear in 2.0