Add Random Missing Data points

btibert

I am sure this is possible, but what is the best way to add missing data to a dataset?  I want add noise and save out the dataset for my class to explore and handle.  

sgenzer

@btibert have you tried the "Add Noise" operator? It's not exactly what you're looking for but I think you could use it to fill the need.

varunm1

Hello @btibert

Is this data related to a general problem or time series problem? If this is a general problem, imputing missing values (operator available) based on an algorithm like KNN is suitable and for time series you can go with replacing missing values operator with mean or replace missing values (series) operator with linear interpolation are suitable.

MartinLiebig

i usually go for generate attribute with:
if(rand()<0.2,MISSING_NUMERICAL,value)

Cheers,
Martin

btibert

Thanks Scott.  I suppose I could get there via multiple splits and declare missing value paths (and then append/union), but good to know about the Noise Operator because I was not aware.  Thanks!

sgenzer

glad we could help @btibert. @mschmitz @varunm1 nice solutions as well. I always love to see how many ways you can tackle a problem in RapidMiner.

Scott