https://bugs.documentfoundation.org/show_bug.cgi?id=94872
--- Comment #8 from [email protected] --- (In reply to Jean-Baptiste Faure from comment #4) > (In reply to thoskk from comment #3) > > > Question from a developer on IRC: what happens if two bars have the same x > > > value? > > xy-scatters don't make much sense if two identical x-values are present. If > > so, the second one belongs to a different x-axis. > > Really? Did you never make chart of jump function? If I'm not completely wrong, even a jump function (or a log function) doesn't give two different y-values for one identical x-value: y = n1 | x > m y = n2 | x <= m (or alike) If you have results from measurements that tell different, the reason is most probably the measuring tolerance of the system. > Please, could you attach a screen copy of a xy (scatter) chart as > bars/columns? Attached two screenshots. First one ("Example for xy-scatter bar (taken from chromatogram simulator") is a good example of how it could look. The program used has more specific options for the task of simulating chromatographic data, but is restricted to 10 entries. > If I understand your request correctly, I think you can do such a chart in > LO: > 1/ assuming x values are in column A and y values in column B, from x-data, > compute an x-step which divides each [x_i ; x_i+1] interval > 2/ create in column C a new x data column with this step > 3/ for each x value in column C available in column A set the corresponding > y value in column D > 4/ for each x value in C not present in column A, set the corresponding y > value in column D to zero > You can use lookup function to do 3 and 4. > 5/ Now you have xy data with constant x step and you can create a bar chart > with columns C and D. Please see second attachment (xy-scattered colums from LO). Not sure if I understood your advice correctly, since I can't see how to assign a separate y-value to each x-value. In addition, how can display of x-axis be set to give reasonable, legible intervals? I'm afraid this method wouln't work, though, since already for this small simulation more than 600 lines of values are necessary. Imagine a simulation of 60 minutes measurement, with increments of 0.001 minutes (totally realistic values for chromatography). This would result in 60,000 lines of values. And besides you would have to enter at least the starting values by hand, LO would maybe soon hit it's limits for some alterations (e.g., when applying functions to simulate different measurement conditions). I made some experiments myself, and the only way to have a similar result as given in attachment 1 is: 1) Enter values by hand (e.g., column A representing the x-values, or minutes, in this case, and column B representing corresponding y-values) 2) Adding two new lines between every line of x-values 3) Calculate an increment for every x, according to your demands (e.g., x-0.01 before and x+0.01 after each x) 4) set corresponding y-values to 0 5) use data to plot xy-scatter graph The result is displayed in the third attachment ("xy-scatter graph"). However, for a series of e.g. 60 minutes measurement with one measuring point every 0.001 minutes, it's very tedious to enter the data. If you copy the values from e.g. an ascii-file, there is no way to automatically enter two new, empty lines after each given line (at least I could not find any function that would allow to do so). This would mean, you had to insert 120,000 new lines by hand. And since selected cells or lines by ctrl-click and shift-click are not handled differently (see M$ Office for reference), one had to click at least 180,000 times. Consider 4 - 5 of these measurement series a day, and you would spend your time doing nothing but clicking. Cheers, Thoskk -- You are receiving this mail because: You are the assignee for the bug.
_______________________________________________ Libreoffice-bugs mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs
