impute missing values

asiddiqasiddiq Member Posts: 25 Contributor II
edited September 2020 in Help
I'm trying to impute missing values for my missing data as illustrated in the below image. However, I'm not sure what the right operator should go inside the impute missing values" operator to make this process work fine. 



  • Options
    BalazsBaranyBalazsBarany Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert Posts: 955 Unicorn

    In Impute Missing Values you use a modeling operator (one in the structure Modeling/Predictive) for imputation. RapidMiner will build a model using the examples where the value is not missing to predict the ones where the value is missing. 

    Select a model that has the capabilities for working with your data. E. g. can it work with nominal data, missings in other attributes, etc. See the Academy videos on the basics of modeling with different models.

    If this is just a routine step to fill missing values with reasonable replacement values, you can go with something like a decision tree, or if you only have numerical attributes, Support Vector Machine, Neuronal Net or Linear Regression. 

    If the replacement is really important and you're relying on it in your analysis, you should cross-validate and optimize different models by taking the part of the data where the target attribute value is not missing. The validation result will help you select the best model.

Sign In or Register to comment.