The Analytics Library functions decisiontree and decisiontreescore handle NULL values differently.
The decisiontree function uses listwise deletion. That is, if a row has a NULL value for any variable (independent or dependent), the function removes that row from the model-building process.
The decisiontreescore function scores all rows except those that have NULL values for the variable a decision node is testing. For example, if a decision node is "age < 50", the function does not score rows that have NULL values for the age variable, but it does score rows that have NULL values for other variables.