Library
What is the recommended imputation policy?
Quote 0 0
Library
Two imputation policies are available:chooseImputation.jpeg [*:1c4bytcv]Choose the Values According to the Law: the values are randomly imputed based on the posterior probability distribution of each variable given the values of all the other variables. Although the variance over the population is preserved, this imputation policy is not optimal at the observation level. You should select this option if the imputation is only an interim step, i.e. the imputed dataset will be used for subsequent learning tasks. [/*:m:1c4bytcv][*:1c4bytcv]Choose the Values with the Maximum Probability: the values are deterministically imputed based on the posterior distribution of each variable given the values of all the other variables. This imputation policy is optimal at the observation level. However, the variance over the population is modified. If this is the final step after the development of a model, this will be the recommended approach.[/*:m:1c4bytcv]Imputation is available for all variables, by selecting Data | Imputation or, alternatively, for an individually selected variable via its contextual menu (in Validation Mode).
Quote 0 0