How to insert "missing examples"?

vwegertvwegert Member Posts: 10 Contributor II
Hello,

I'm looking for a way to generate "missing examples" into an example set. Basically, I have and example set with a date, a (polynomial) key and some numeric attribute that represents a measured value. In my process, I'm splitting this example set into partitions defined by the key (I'm using the Loop Values operator) and produce a Scatter plot for each value over time. While this works rather nicely, there's a slight problem left: Since the selection of stuff that is measured changes all the time, I don't get a consistent date range for every key value. For example, someone might decide that key = A is no longer worth measuring and turn it off, probably enabling key = B instead. This is not a problem for the process itself or the plot output, it all works as designed - but it's a problem for the user because the time axes of the resulting plots differ, sometimes wildly. Since the plots are shown directly beneath each other, the user will be tempted to compare the measurement data at a certain point in time, probably missing the fact that at the same (graphical) x position a different date value is plotted for every graph.

My question: What is the easiest way to ensure that all of the plots scale the time? I haven't been able to use the Scatter range list (of the Report operator) for this - maybe because I don't fully understand what the correct values for a date range might be. As an alternative, would it be possible to generate additional examples for the missing date/key combinations that have an empty numeric attribute and still have the Scatter plot display that date on its axis?

Thanks
  Volker

Answers

  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,503 RM Data Scientist
    Hi Volker,

    i am not a hundred percent sure. But can't you create an exampleset which has the correct axis, join that with yours and then fill the missings?

    Best,
    Martin
    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
Sign In or Register to comment.