I am understanding this more and more and appreciate the wonderful feedback.
I have played around a little with this example and found several interesting things which are either helping or confusing me more, sometimes I can't tell which. 🙂
A- When I plotted this using the line chart it did indeed come out the way I expected. As you may note one of the things I did is change the dates a little (the last date is different for the 2 stocks) and I got the appropriate "shorter" line on the one that ended sooner. So far, that's good because it's what I expected.
B- I then made another change. I changed the expression from sum(PRICE) to simply say PRICE. When I did this there appears to be no difference in the chart. Therefore, it makes me wonder why the sum() calculation is needed at all?
C- I then removed the 2nd dimension (STOCK_NAME) but the resulting graph was very peculiar: it seemed to "throw away" almost all of the data and only seemed to pick 2 data points to display. I have no idea why that looks that way.
D- I then changed the expression back to sum(PRICE) and "all of the points" seemed to come back but this time they are (I believe understandably) now "merged together" (i.e. summed) and then plotted. This is the "problem" I believe I am having with my data (data merged together) even when I select a 2nd dimension. This makes me believe that perhaps the 2nd dimension I am choosing isn't right. I have tried selecting several of the other values which might be appropriate but none of them seem to have any effect at all; the graph has no change after I add the dimension and click "Apply" to see it. Let me modify that slightly -- I am not getting what I want when I select the specific item (i.e. the name) as the 2nd dimension, however I have been able to get multiple lines when I select certain other columns of the data.
E- I was getting the "multiple lines" for some of the fields (I haven't figured out the pattern) when I had only one expression (the minimum value.) When I then added in the expression for the maximum value "suddenly" the legend on the top right corner changes from the 2nd item mentioned in the dimensions into the sum(min) and sum(max) and the 2nd dimension item simply "disappears."
All of the above is for the "1st chart" both because it seems to be simpler (and it seems as though I need to keep things simple right now) and also because I do not want to have to "call out" each individual name in the expression field because there isn't any way I am going to know what they are in a production environment.
At this point I guess my understanding would be helped if I understood why in #E above adding the 2nd expression seems to get rid of the 2nd dimension.
If you like I believe I could probably upload an XL file of a chunk of my data because I have a "table chart" which shows this subset of information I'm interested in. (This gets rid of the problem of having a lot of other extraneous columns and data.) As I look at this some more perhaps it may be necessary for me to create a "virtual field" (which is the impression I get from your 2nd example) which combines the timestamp and the value but my head still doesn't understand how that would make it possible to get my desired results. Perhaps if I understood why #C above behaves the way it does I might understand better how to get the data into the chart correctly.
I am confident QlikView is going to be able to display the chart I want once I figure out how to specify it appropriately. I continue to be surprised that none of the examples I have found so far show stock market data from multiple stocks over a period of time similar to what the 1st example chart showed.
B - The sum() is NOT needed at all in this particular chart because there is only a single value of PRICE for any combination of the given dimensions. I'd probably personally use only(PRICE) which will also return the same results. I think the reason I do it is based on SQL concepts. The dimensions of the chart are your group by. SQL requires an aggregation function if you're doing a group by. So I use an aggregation function, even though QlikView doesn't require one in this case. Feel free to not do so, as I can't think of any QlikView-centric reason for using only() here.
C & D - Here, a sum() or something else IS needed. Your PRICE is no longer unique based on the date dimension. So for 02/01/2010, you have a price of 200 and a price of 30. What should it display? It doesn't know, so it displays nothing. Only two data points have a unique price for the date, so only those two points display.
E - As I said earlier, you can have a second dimension or a second expression, but not both. Well, technically you can have both, but it won't give you lines for both. It appears to ignore your second dimension in that case. Why? Well, honestly, I wouldn't have designed it that way, so I can't really say why. I'd probably have moved the second dimension to the X axis, giving you a separate line chart for each, though I don't think that's what you personally want. The other reasonable alternative in my mind would be to automatically concatenate the second dimension and the expression name, and so overlap all of the lines instead of giving you "separate" line charts. I think that's what you're looking for. QlikTech apparently chose to do nothing with it instead. Consider it an unfortunate limitation, then.
I see your Excel file, but I'm not sure what you want your resulting chart to look like, so it's only half of being able to give you a solution. Can you mock it up or otherwise describe it?
Oh! Oh! Oh! You hit the nail on the head. The key phrase I did not understand was "you can have a second dimension or a second expression, but not both." It also helped a lot to know that, once QlikView couldn't "use" a dimension it was simply ignoring it. Perhaps a "message" of some kind would be appropriate so you'd know it's not helping...
When I removed the 2nd expression I got the chart I was looking for! I am attaching a picture of what I got (after adding in a few more TTagID items.) It's giving exactly what I want to see - the "trend" of each item following a "pattern" but some much higher or lower than others.
I'm next going to see about using a "stock" like chart (I realize I'll have to convert to Combo to do that, so it's not trivial but I'll go do it shortly) and try using the max/min feature there to get both values on one line for each tag. Then again, seeing how "crowded" it can get with only a few tags, maybe separate min and max charts makes more sense. (Pretty easy to do also, get the chart all set up for max and then clone it and change the expression to min - cool!)
Thanks also for the insight on sum() and your use of it. I understand perfectly being "used to" a certain way of doing things and keeping that up even if you don't have to.
As to the C&D comment "It doesn't know, so it displays nothing" that's interesting too. Maybe it's simply my newness to this, but simply "ignoring" a problem like this doesn't help me learn what to do to fix the problem. I get "errors" in Visual Studio many times although it plows on ahead and gives me a report. When it doesn't look right, I go to the list of error messages first to get an idea what might be wrong. (It's not always the problem I'm looking for, but it is a problem...) I guess QlikView doesn't have a way to display "errors" like this. Perhaps I'll get used to it...
Thanks again for the great advice. I appreciate the time and effort and hope to get good enough to help contribute to the community when I get a little better.
Bottom line for this problem: "you can have a second dimension or a second expression, but not both"
GSaxer01 wrote:As to the C&D comment "It doesn't know, so it displays nothing" that's interesting too. Maybe it's simply my newness to this, but simply "ignoring" a problem like this doesn't help me learn what to do to fix the problem. I get "errors" in Visual Studio many times although it plows on ahead and gives me a report. When it doesn't look right, I go to the list of error messages first to get an idea what might be wrong. (It's not always the problem I'm looking for, but it is a problem...) I guess QlikView doesn't have a way to display "errors" like this. Perhaps I'll get used to it...
To me, it's a consequence of QlikView being "looser" than SQL. SQL would REQUIRE that you specify an aggregation function. It requires you to say what to do with multiple values if they're possible. QlikView doesn't require any such thing; it just returns null(). I'm sure it could return an error, as you'll sometimes see error messages instead of your chart, so QlikTech probably decided that this wasn't an error, and that null() was the correct response. I don't really agree, but I'm used to it. As with all new "programming languages", you get used to the quirks eventually.
