PROJECT 94/2204 "METHODS FOR HANDLING ILL-CONDITIONED DATA IN THE FSU"

  • S U M M A R Y
  • The Project Information
  • Multiple imputation method for binary and polytomous data models

    Implementation and testing

     

     

    Due to the INTAS 94/2204 Project, the CENTER FOR HANDLING ILL-CONDITIONED DATA (CHILD, for short), has been established at the Department of Math. Modeling in Ecology and Medicine of CCAS. The Gibbs algorithm was recently added to the Missing Data Machine of the CHILD. The algorithm can be used to impute missing values in continues variables. You can send us your data with missing values to shakin@ccas.ru and/or to child@tradition.ru and we impute the values and send them back to you. A format of the data must be:

    0|13.141607|10.625606|*|3.6264823

    *| 16.58347|17.382634|12.277927|12.337908

    1| 11.04347|*|11.397927|13.417908

    *| 10.63| 9.04| 10.5| 8.79

    1| 4.92| 5.25| *| 2.75

    1| *|12.052634|9.7779267|14.327908

    0| 6.38| 4.79| 6.46| 10.88

    0| *| 11.29| 9.38| 8.71

    *| 11.04347|8.0126337|9.4879267|9.6279082

    ......................................................................

    where “*” means missing value and “|” is delimiter. Also we need a discription what type are variables (continues, binary, …) Any comments on the underlying problem and related data generation model are always appreciated.