I have a problem with the Random Boolean Assigner node. When I set the number of TRUE values using the method called “Use number of true values” it generate more TRUE values than those expected. As you can see in the attached example, I set 10 values but I obtain 14.
Can somebody confirm me if it is a bug or if I misunderstood something?
Thanks in advance,
I can confirm this bug. And just for fun, I tried to find some quick alternatives.
Thank you for your reply. It seems we're tuned as I had generated a workaround like your second one!
To what deals with the bug it would be good if some KNIME developer becomes aware of this problem so they can fix it for the next release.
I am sorry but this is not a bug, this is the intended behaviour. It gets even better, you will get a different number of true values if you change your seed.
In the case event number is selected, the probability is set to eventnumber/rownumber. This feature is for simple configuration possibilities.
If you want to have this standard behaviour: Create a table with two rows: true 100, false 1000 and use the one row to many and maybe a shuffling.
Then maybe the naming isn't the best, right? Also, while you're changing the name, why not add this function for a future release?
More basically, I may be wrong, but the current method seems to assume some normal distribution of the true count. Which would mean there should maybe be a way to specify the variance of that distribution. If there were, setting it to zero would have solved this issue as well.
Thank you for your clarification. Now is fine with me.
In my opinion it would be good to clarify this also in the node description. Maybe with a more explicit sentence. Thanks again.