Permutation Tests of NELS:88 Missing Data

(make-array (list l1 l1) :initial-element 0)) (ra (mapcar #'(lambda (x y) (mapcar #'(lambda (z w) (setf (aref va x z) (aref smm-1 y w))) missind (iseq (length missind)))) missind (iseq (length missind))))) va)) 22) (list vi smm-1 smo))) (t (cond ((not (not missind)) (list (inverse sig))) (t (list 0 (mapcar #'(lambda (x) (elt x i)) data))))))) (defun impute (params missind presind mu data i) "Args: (PARAMS MISSIND PRESIND MU DATA I). Imputes the missing data vector for case i indexed by missind by regressing the missing values on the non-missing values. PARAMS is a list of matricies and lists. MISSIND are the indicies of the missing variables for case i. PRESIND are the indicies of the present variables for case i. MU is the current mean vector. DATA is a column list of the data. This funtion is only called if there are at least one present variable and at least one missing variable for case i. Returns the list of imputed and observed values for case i." (let* ((dati (mapcar #'(lambda (x) (elt x i)) data)) (newvals (+ (select mu missind) (inner-product (matmult (second params) (third params)) (-(select dati presind) (select mu presind)))))) (setf (select dati missind) newvals) dati)) (defun make-vi (smm-1 missind l1) "Args: smm-1 missind l1. Returns a matrix of size l1xl1 with smm-1 in place of the missing variables and zeros elsewhere." (let* (21 (setf results (list data mu (inverse siginv) (butlast likehist))) (print-results results))) (defun print-results (results) #| (format t "The Imputed Data Matrix is: ~%") (print-matrix (first results)) |# (format t "~% The Estimated Variable means are: ~%") (print (second results)) (format t "~%~% The Estimated Covariance Matrix is: ~%") (print-matrix (third results)) (format t "~%~% The Log-Likelihood History is: ~%") (print (elt results 3)) (format t "~% All of the above are in the variable results ~%") (def pp (plot-points (iseq (length (combine (elt results 3)))) (combine (elt results 3))))) (defun param-est (sig data missind presind mu i torf l1) "Args: (SIGMA DATA MISSIND PRESIND MU I TORF L1) Takes in the data, missing and present indicies and estimates the parameters of a multivariate normal distribution. SIG, the inverse covariance matrix should be in the form of a matrix. Data must be in the form of a list of vectors MISSIND must be in the form of a list for case i. PRESIND must also be in the form of a list for …