topic Re: Handling time changes when graphing existing data in LabVIEW

Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 10:47:53 GMT

My challenge is to graph existing temperature and event data from several files in one graph. On the surface, this is fairly easy. I've been able to use a XY graph to plot the data and add cursors where events occur. If only life were so simpe.

Our main (non-labview) system writes the CSV files using what we call an event sequence index. This is a sequential number that is the same in each file and ties the records together in case the user decides to change the time or date, which they do. Users are a pain like that. 🙂 So, I use it to tie everything together, build a bunch of arrays using the time tied to the event sequence number in what we condsider the main CSV file and graph the data. The problem is, the Labview XY graph seems entirely time based and I can't graph the data in a strip chart form, at least as far as I can figure out. If data is graphed that appears older than some, but really isn't, it is drawn to the left in date/time order. I haven't tried yet to see what happens if there is data for the same date and time. that should be fun.

I'm probably missing something simple, at least I hope.

Any ideas would be appreciated!

Re: Handling time changes when graphing existing data

GerdW — Fri, 14 Oct 2016 10:58:03 GMT

Hi Steve,

a XY graph plots data as you provide it. When a data point is to the left of another point then its X value is smaller (using a default X axis with bigger values to the right).

Maybe you need to check your data? Or even better: attach a VI with some meaningful default data inside!

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 11:23:33 GMT

Thanks for responding.

I'm graphing using the time stamp from the primary CSV file tied to the Event Sequence ID I'm using to tie everything together. This is my "X" value. The graph plots data based on time in this case and not by where it is in the array, at least the way I've managed it.

I guess I could use my Event Sequence ID as my X value. That would solve the issue but create two new problems. The first is that the Event Sequence ID means nothing to a user, so I'd have to figure out a way to hide it and display the corresponding date/time label in it's place, which seems difficult. The other problem is that each data point is not linear. Most are 1 minute apart, but if events occur, they may only be a second or two apart. That may cause the graph to look a bit funky.

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 12:52:24 GMT

It occurs to me that my words may to confusing. I have that effect some times. So, I've attached an example of my problem. In this example, I have an array of clusters that include a "time" stamp and the associated data for each. I've sequenced the first five"time" elements from 5 to 9. I then made the sixth element 1. This is the same issue I'm running into when I use a true time stamp. In this example, I need to somehow make the "1" time with the data at "8" appear to the right of the "9" time stamp.

Re: Handling time changes when graphing existing data

Bob_Schor — Fri, 14 Oct 2016 12:58:18 GMT

If you want to plot a single array of equally-spaced elements, use a Chart. If you want to plot X,Y pairs, which plot as X,Y (so if X goes 5, 6, 7, 8, 1, it plots the last point to the left of the other four), use a Graph. If you want to plot X,Y pairs as though they were equally spaced points, ignoring the X values, either plot the Y points as a Chart or replace the X points with 1, 2, 3, 4, ... .

Bob Schor

Re: Handling time changes when graphing existing data

billko — Fri, 14 Oct 2016 13:15:10 GMT

I guess my suggestion would be to decide which of the plots is going to be your reference and use those timestamps for all plots. The fact that the customer wants a unique number for each piece of data indicates that they expect the data to be taken at the same time. Since no timestamp is exactly accurate, I don't consider this "fudging" data. it is merely a method of aligning data.

Re: Handling time changes when graphing existing data

billko — Fri, 14 Oct 2016 13:17:31 GMT

Alternatively just use the reference number for your chart and just use the timestamps as references. That would probably be the best way to go, actually.

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 13:33:44 GMT

Billko,

I agree. I could use the Event Sequence ID (reference as you called it) and the main problem is solved. I would just need to figure out how to make the timestamps visible to the user after that. I suppose I could use cursors for that. The problem then becomes the spacing of those cursors and what to do as the user zooms in and out. It also creates a problem with regard to the non-linear timing of my samples, but I think we can live with that.

Bob,

Thanks for the reply. I'm just having a hard time determining how I would chart it versus graphing it. I'm dense like that. I'll keep playing around.

Thanks all!

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 13:35:25 GMT

Oops. I accidently marked it as solved. Duh...

Re: Handling time changes when graphing existing data

billko — Fri, 14 Oct 2016 13:36:51 GMT

You can unmark it. 😞

😄

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 13:51:46 GMT

Re: Handling time changes when graphing existing data

billko — Fri, 14 Oct 2016 13:51:51 GMT

You can use the event structure to capture the cursor release and use that to index the timestamp and show it as either a popup or as annotation.

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 13:54:22 GMT

Another ignorant question. What's the best way to get the time stamp to appear on the graph in the first place since my event sequence number (reference number) is now the X axis label? Can you add text to a cursor?

Thanks for your help so far.

Re: Handling time changes when graphing existing data

billko — Fri, 14 Oct 2016 14:03:07 GMT

@SteveC_43 wrote:
Another ignorant question. What's the best way to get the time stamp to appear on the graph in the first place since my event sequence number (reference number) is now the X axis label? Can you add text to a cursor?

Thanks for your help so far.

You can change the cursor name after you let the cursor go. Just make sure to make the cursor mode anything other than "free".

Re: Handling time changes when graphing existing data

billko — Fri, 14 Oct 2016 14:08:31 GMT

Actually using an annotation is better - changing the name of the plot will be problematic if you ever want to reference the plot by name (which is almost certain).

I have no idea what i was thinking in the post above - yeesh.

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 14:53:32 GMT

Yeah, I'll play around with that. Thanks for your help!

Re: Handling time changes when graphing existing data

drjdpowell — Fri, 14 Oct 2016 20:21:50 GMT

You could avoid the problem entirely if instead of a "sequence index" and a User-made-up pretend time, you recorded the actual time.

Re: Handling time changes when graphing existing data

billko — Fri, 14 Oct 2016 20:30:25 GMT

@drjdpowell wrote:
You could avoid the problem entirely if instead of a "sequence index" and a User-made-up pretend time, you recorded the actual time.

Except for those darned, pesky customer requirements. 😄

Re: Handling time changes when graphing existing data

SteveC_43 — Fri, 14 Oct 2016 23:20:39 GMT

It's never a pretend time. Ever heard of daylight savings? How about a system we send to Europe with existing data and they want to change a time or date? The only way to tie everything together is to NOT do it by date/time or it get's incredibly messy. Otherwise you get duplicate times and you don't know which time is the correct one to tie events to.

That system, which creates the files I'm reading, was written by very good programers with years of experience writing code for stuff you may use (or drive) every day. They don't just make stuff up or pretend.

Maybe you were just being humorous. If so, just ignore me. 🙂

Re: Handling time changes when graphing existing data

billko — Sat, 15 Oct 2016 01:54:54 GMT

@SteveC_43 wrote:
It's never a pretend time. Ever heard of daylight savings? How about a system we send to Europe with existing data and they want to change a time or date? The only way to tie everything together is to NOT do it by date/time or it get's incredibly messy. Otherwise you get duplicate times and you don't know which time is the correct one to tie events to.

That system, which creates the files I'm reading, was written by very good programers with years of experience writing code for stuff you may use (or drive) every day. They don't just make stuff up or pretend.

Maybe you were just being humorous. If so, just ignore me. 🙂

Your first post implied a whole lot by describing everything with minute detail. It seemed that there was already a well established "way of doing things", if not actually established by customer requirement.

I'm sure that @drjdpowell didn't pick up on that. That's why I had the big smiley at the end. 🙂