Practical Predictive Analytics

Missing at Random (MAR)

Missing at random (MAR) implies that the missingness of a value is related to another variable in the analysis. It is somewhat of a misnomer, since there is nothing random about it.

So, another question to ask when dealing with missing values is "Is the fact that there are missing values in this variable related to the dynamics of any other variable?" Again, using our survey analogy as an example, missing values for, "What is your Expected Salary?" might occur for certain demographic groups who did not answer the question. If you found a statistically significant number of missing values at one level of a factor, but not as many within another level, that would suggest that the data was missing at random. A variable ...