creating non aggregated graphs

[#1]

l.s

i have been using qlik quite some time now and a happy user. Qlik is very helpful in learning from historical data. What i want to do now is also show realtime data. we have sensors in the field at our customers and they produce very much data. I want to show this data to monitor the health of the sensors. So i have a timestamp when the data was collected from the sensor and have the diagnostic and process data at that timestamp. Now i want to show this in a graph without aggregating the data. So lets say the timestamp on the x-as and the measured values on the y-ax. But the only option that i have is to count, sum, average etc.

That is not what i need. I need to visualise the data that is measured without aggregation. Can someone help me?

[Anonymous]

dont use any aggregate function like Count, sum, avg()...

Simply use measure field as a expression.

Say, Amount is your measure then don't use sum(Amount), simply take Amount as a expression

[reddy-s]

Hi Eelco van,

You can use any measure field with out aggregating it as a measure. Just make sure its a measure. That's It!

[simondachstr]

This is not entirely correct. If you use your measure as a field expression Qlik will implicitly use the only function and return distinct values. Hence, if a dimension has multiple measures of the same value, it will only be returned once. If you truly want to see values in a chart/table on its most granular level you will need to create a RowId() field in the script and use this as your dimension/measure.

[JonnyPoole]

I agree with martin,  qlik visualizations typically represent some sort of aggregation. You can avoid that by introducing a unique key (different value for each row of data) and then including that key in the chart.   Only() and concat() functions will render those values without doing a sum or max etc...    The key again is to be including a dimension that is unique across the measure values. 

If you had an example visual of what you want to show (like if you have 2 measure values for 1 dimension value) and how you'd like to visualize it , that would help make suggestions.

[#6]

Hi Guy's

thnx all very much for your supplies, @martin and Jonathan you guy's understand what i mean. However i'm having difficulties creating this my self. So have taken some sample of the raw data and included that.

What i would like to have in a graph on the horizontal axis is the time stamp and on the vertical axis the three remaining collums. Could you give me the code on how to load this in qlik so that i can create this graph?

thnx in advance!

[simondachstr · Accepted solution]

Hi Eelco,

This will not be easy to achieve as all three metrics (columns in your example that is) have different magnitudes. You will be able to achieve to display all three columns in a chart (a line chart most likely, as you are interested in the change over time), but the scaling will not allow you to see much.

Depends what you are interested to see here. If you want to see each column by itself - take the above approach and make the respective selections on your filters. If you want to see the change over time of all three columns in relation to each other, I recommend normalizing them by the first value of the series.

Technically, just load the table as is with the QlikView wizard (should be easy to achieve). Choose the line chart, add Timestamp as the dimension, and the other three columns as measures (no aggregations required here).

[#8]

Hi Martin,

thnx thats it! i never knew it was this easy, just removing the aggretion in the formula. Thnx!

[#9]

Hi Martin,

i have just one question, earlier in the discussion you mentioned that i should use RowId() to make sure that i see all the values. My question to you is, do you have to use this when you load the data or in the expression that you use for making a graph.

RowId([[CH05 SAMPLE COUNT]] Current Output C])

This for example in the expression for creating a graph results in a flatline.

[Anonymous]

HI mahlerma‌, Are you able to explain how to create a non aggregated measure in general? We have data that is in percents.