It seems I can only make time series (including with TSBV) of numerical values. What I have are gateways that send a finite set of event types. What I would like to do is plot in a time series which type event was sent. Preferably without full bars but just points like the 'heat' map.
Is this possible, or must the y-axis always be a numerical value? I can make the x-axis show the set of event types, but not the y-axis.
Its not really a time series but on the x-axis I plot the time of the event, and if a full bar must be used (okay) the value of the bar would be up to the event type (one of perhaps 10 types).
Is there another visualization that is more appropriate? I could not use a table either, as that had the same restriction.
I tried such an option, however on the y-axis I could not select a non-numeric field as a variable. On the x-axis I was able to plot the time of the event, but the y-axis was forced to be some count, sum, average, unique count, etc. which all required numeric fields and were the only fields offered.
Was looking for a y-axis that might be labeled as the event type and a 'dot' or 'circle' would be placed on the y-value given that event type as a function of the event time. One could then see what event type happened at what time and drill further down form there (say one was debugging some problem).
In the Lens editor you have an Horizontal vs Vertical configuration based on ordinal values, then a Cell value for the specific metric value which colors the cell.
For instance this is the Vertical axis configuration I see:
I did not have to use the Lens editor. Turns out I could make x and y axes in the buckets selection without invoking the metric count. I will need to check if I can do that in all the other visualizations. Would be great if I could!
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.