LV 2009 error bar plots suggestions

[X.]

I have mentioned my mixed feelings about the new error bar plot introduced in 2009 on the info-labview list and have been asked to move this discussion to this forum.

So there it goes:

Introducing the possibility to natively display plots with error bars associated with each data point is a nice touch that brings LV one step closer to being a full-fledged data analysis and representation environment. However, the feature is limited in allowing only a single axis error bar (and on the Y axis only). This might come from NI's hardware bias (time recording of waveforms, in which the time axis does not have any "error" to it, being clock driven). Some of us use LV for a tad more than just recording waveforms and displaying them. We use complex instruments, process our data extensively, and when times come to plot a XY graph (and this is true for higher dimensionality data too), each coordinate (X, Y, etc) of each data point may be affected by a statistical or experimental uncertainty.

In this kind of situation, each data point should be represented as a center point with two orthogonal error bars (with different sizes, and possibly, as offered in the new error plot, different plus and minus error bars). In most cases, the plus and minus error bars will be identical (representing the standard deviation of the data), so it would be advantageous to have an option that allows specifying that and avoid passing twice the same array of error bars to the plot formatting VI.

Note that introducing this feature wouldautomatically bring up the request to have fitting algorithms that take this kind of dual axis error bar representation into account (which most data analysis software do nowaday).

I hope this brief suggestion will be supported by other fellow LabViewers...

X.

[Albert.Geven]

I agree completely and I hope that more graph functionality comes available for all axes.

Another missing property is a scrollbar for the Y-axis.

[LewDrake]

This error bar plotting functionality can also be useful for displaying data where each X-axis point represents a block of time, over which some Y value variation occurs (for example, stock values vs time).  For stock value plots, the a common type of plot is called a "candle chart".  See:

http://stockcharts.com/school/doku.php?id=chart_school:chart_analysis:introduction_to_candlesticks 

for an explanation.  In a single glance, the candle plot shows the data's range and opening and closing values.  The candle chart uses coloration to indicate the opening and closing values.

I suggest that the developers keep this capability in mind when enhancing the functionality of the error plotting function.

[GregSands]

The candle chart as described is mostly only useful for stock information.  For statistical data analysis, the box-and-whiskers plot is much more useful, which adds a line in the middle of the box.  The various features typically indicate the min, max, and 25/50/75 % percentiles (quartiles) of the data (50% being the median).

[Robbob]

All,

These are great suggestions, and we always enjoy getting feedback from real-world situations where LabVIEW could help out more.  I really think this should be thrown on the LabVIEW Idea Exchange. Our development team looks over all the suggestions there, and it would have much more visibility than just being posted in this thread.

Also, while you are there, be sure to vote on ideas that you like!  Some of the top ideas are so obvious, you will be surprised they weren't implemented already!