Fix ROC behaviour in the presence of missing values.

Make sure that the ROC command's behaviour is correct,
when missing values appear in the result variable.

diff --git a/src/language/stats/roc.c b/src/language/stats/roc.c
index e6251b36b38ff79c495931e1b805f4489df3c992..0709992ce469ae113814f6996b549fe0d684b7fc 100644 (file)
--- a/src/language/stats/roc.c
+++ b/src/language/stats/roc.c
@@ -47,6 +47,7 @@ struct cmd_roc
 {
   size_t n_vars;
   const struct variable **vars;
+  const struct dictionary *dict;
 
   struct variable *state_var ;
   union value state_value;
@@ -66,6 +67,11 @@ struct cmd_roc
 
   bool invert ; /* True iff a smaller test result variable indicates
                   a positive result */
+
+  double pos;
+  double neg;
+  double pos_weighted;
+  double neg_weighted;
 };
 
 static int run_roc (struct dataset *ds, struct cmd_roc *roc);
@@ -86,6 +92,9 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
   roc.ci = 95;
   roc.bi_neg_exp = false;
   roc.invert = false;
+  roc.pos = roc.pos_weighted = 0;
+  roc.neg = roc.neg_weighted = 0;
+  roc.dict = dataset_dict (ds);
 
   if (!parse_variables_const (lexer, dict, &roc.vars, &roc.n_vars,
                              PV_APPEND | PV_NO_DUPLICATE | PV_NUMERIC))
@@ -314,10 +323,25 @@ static bool
 match_positives (const struct ccase *c, void *aux)
 {
   struct cmd_roc *roc = aux;
+  const struct variable *wv = dict_get_weight (roc->dict);
+  const double weight = wv ? case_data (c, wv)->f : 1.0;
 
-  return 0 == value_compare_3way (case_data (c, roc->state_var),
+  bool positive = ( 0 == value_compare_3way (case_data (c, roc->state_var),
                                  &roc->state_value,
-                                 var_get_width (roc->state_var));
+                                            var_get_width (roc->state_var)));
+
+  if ( positive )
+    {
+      roc->pos++;
+      roc->pos_weighted += weight;
+    }
+  else
+    {
+      roc->neg++;
+      roc->neg_weighted += weight;
+    }
+
+  return positive;
 }
 
 
@@ -407,10 +431,10 @@ process_group (const struct variable *var, struct casereader *reader,
               struct casereader **cutpoint_rdr, 
               bool (*pos_cond) (double, double),
               int true_index,
-              int false_index
-              )
+              int false_index)
 {
   const struct variable *w = dict_get_weight (dict);
+
   struct casereader *r1 =
     casereader_create_distinct (sort_execute_1var (reader, var), var, w);
 
@@ -570,7 +594,11 @@ prepare_cutpoints (struct cmd_roc *roc, struct roc_state *rs, struct casereader
     {
       for (i = 0 ; i < roc->n_vars; ++i)
        {
-         const double result = case_data (c, roc->vars[i])->f;
+         const union value *v = case_data (c, roc->vars[i]); 
+         const double result = v->f;
+
+         if ( mv_is_value_missing (var_get_missing_values (roc->vars[i]), v, roc->exclude))
+           continue;
 
          minimize (&rs[i].min, result);
          maximize (&rs[i].max, result);
@@ -598,14 +626,12 @@ prepare_cutpoints (struct cmd_roc *roc, struct roc_state *rs, struct casereader
 }
 
 static void
-do_roc (struct cmd_roc *roc, struct casereader *input, struct dictionary *dict)
+do_roc (struct cmd_roc *roc, struct casereader *reader, struct dictionary *dict)
 {
   int i;
 
   struct roc_state *rs = xcalloc (roc->n_vars, sizeof *rs);
 
-  struct casewriter *neg_wtr = autopaging_writer_create (casereader_get_proto (input));
-
   struct casereader *negatives = NULL;
   struct casereader *positives = NULL;
 
@@ -614,6 +640,15 @@ do_roc (struct cmd_roc *roc, struct casereader *input, struct dictionary *dict)
   struct subcase up_ordering;
   struct subcase down_ordering;
 
+  struct casereader *input = casereader_create_filter_missing (reader,
+                                                              roc->vars, roc->n_vars,
+                                                              roc->exclude,
+                                                              NULL,
+                                                              NULL);
+
+
+  struct casewriter *neg_wtr = autopaging_writer_create (casereader_get_proto (input));
+
   prepare_cutpoints (roc, rs, input);
 
   positives = 
@@ -948,13 +983,11 @@ show_summary (const struct cmd_roc *roc)
   tab_text (tbl, 0, 3, TAB_LEFT, _("Negative"));
 
 
-#if 0
   tab_double (tbl, 1, 2, 0, roc->pos, &F_8_0);
   tab_double (tbl, 1, 3, 0, roc->neg, &F_8_0);
 
   tab_double (tbl, 2, 2, 0, roc->pos_weighted, 0);
   tab_double (tbl, 2, 3, 0, roc->neg_weighted, 0);
-#endif
 
   tab_submit (tbl);
 }

diff --git a/tests/command/roc.sh b/tests/command/roc.sh
index 263d399e64b8b9297cad668a08a6a4b4c5cd45a8..57dad752fa73c705781171f12fd31a0845b07b16 100755 (executable)
--- a/tests/command/roc.sh
+++ b/tests/command/roc.sh
@@ -68,6 +68,7 @@ begin data.
 2 3 4  1
 2 4 14 0
 3 5 10 1
+. . 1  0
 3 1 5  0
 4 2 14 1
 4 3 2  0

diff --git a/tests/command/roc2.sh b/tests/command/roc2.sh
index 693fa831c0f640430bb1d18c8dbbb08839688fe9..650db52fd4ce0d5bc0780562467354a361ad03e6 100755 (executable)
--- a/tests/command/roc2.sh
+++ b/tests/command/roc2.sh
@@ -108,6 +108,15 @@ if [ $? -ne 0 ] ; then no_result ; fi
 activity="compare results"
 perl -pi -e 's/^\s*$//g' $TEMPDIR/pspp.list
 diff -b  $TEMPDIR/pspp.list - << EOF
+1.1 ROC.  Case Summary
+#========#===================#
+#        # Valid N (listwise)#
+#        #==========#========#
+#a       #Unweighted|Weighted#
+#========#==========#========#
+#Positive#        14|  14.000#
+#Negative#        14|  14.000#
+#========#==========#========#
 1.2 ROC.  Area Under the Curve (x)
 #====#==========#===============#=======================#
 #    |          |               | Asymp. 95% Confidence #