Fix cleanup of ROC command.

[pspp-builds.git] / src / language / stats / roc.c

diff --git a/src/language/stats/roc.c b/src/language/stats/roc.c
index e6251b36b38ff79c495931e1b805f4489df3c992..e7dec56b8f1a498c61d59139cdeec5f65abd22b6 100644 (file)
--- a/src/language/stats/roc.c
+++ b/src/language/stats/roc.c
@@ -16,10 +16,10 @@
 
 #include <config.h>
 
-#include "roc.h"
 #include <data/procedure.h>
 #include <language/lexer/variable-parser.h>
 #include <language/lexer/value-parser.h>
+#include <language/command.h>
 #include <language/lexer/lexer.h>
 
 #include <data/casegrouper.h>
@@ -47,8 +47,9 @@ struct cmd_roc
 {
   size_t n_vars;
   const struct variable **vars;
+  const struct dictionary *dict;
 
-  struct variable *state_var ;
+  const struct variable *state_var ;
   union value state_value;
 
   /* Plot the roc curve */
@@ -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);
@@ -86,21 +92,24 @@ 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))
-    return 2;
+    goto error;;
 
   if ( ! lex_force_match (lexer, T_BY))
     {
-      return 2;
+      goto error;;
     }
 
   roc.state_var = parse_variable (lexer, dict);
 
   if ( !lex_force_match (lexer, '('))
     {
-      return 2;
+      goto error;;
     }
 
   parse_value (lexer, &roc.state_value, var_get_width (roc.state_var));
@@ -108,7 +117,7 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
 
   if ( !lex_force_match (lexer, ')'))
     {
-      return 2;
+      goto error;;
     }
 
 
@@ -131,7 +140,7 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
              else
                {
                   lex_error (lexer, NULL);
-                 return 2;
+                 goto error;;
                }
            }
        }
@@ -155,7 +164,7 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
          else
            {
              lex_error (lexer, NULL);
-             return 2;
+             goto error;;
            }
        }
       else if (lex_match_id (lexer, "PRINT"))
@@ -174,7 +183,7 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
              else
                {
                  lex_error (lexer, NULL);
-                 return 2;
+                 goto error;;
                }
            }
        }
@@ -197,7 +206,7 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
                  else
                    {
                      lex_error (lexer, NULL);
-                     return 2;
+                     goto error;;
                    }
                  lex_force_match (lexer, ')');
                }
@@ -215,7 +224,7 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
                  else
                    {
                      lex_error (lexer, NULL);
-                     return 2;
+                     goto error;;
                    }
                  lex_force_match (lexer, ')');
                }
@@ -241,14 +250,14 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
                  else
                    {
                      lex_error (lexer, NULL);
-                     return 2;
+                     goto error;;
                    }
                  lex_force_match (lexer, ')');
                }
              else
                {
                  lex_error (lexer, NULL);
-                 return 2;
+                 goto error;;
                }
            }
        }
@@ -259,9 +268,14 @@ cmd_roc (struct lexer *lexer, struct dataset *ds)
        }
     }
 
-  run_roc (ds, &roc);
+  if ( ! run_roc (ds, &roc)) 
+    goto error;;
+
+  return CMD_SUCCESS;
 
-  return 1;
+ error:
+  free (roc.vars);
+  return CMD_FAILURE;
 }
 
 
@@ -314,10 +328,25 @@ 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 +436,10 @@ 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 +599,11 @@ 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 +631,12 @@ 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 +645,22 @@ do_roc (struct cmd_roc *roc, struct casereader *input, struct dictionary *dict)
   struct subcase up_ordering;
   struct subcase down_ordering;
 
+  struct casewriter *neg_wtr = NULL;
+
+  struct casereader *input = casereader_create_filter_missing (reader,
+                                                              roc->vars, roc->n_vars,
+                                                              roc->exclude,
+                                                              NULL,
+                                                              NULL);
+
+  input = casereader_create_filter_missing (input,
+                                           &roc->state_var, 1,
+                                           roc->exclude,
+                                           NULL,
+                                           NULL);
+
+  neg_wtr = autopaging_writer_create (casereader_get_proto (input));
+
   prepare_cutpoints (roc, rs, input);
 
   positives = 
@@ -948,13 +995,11 @@ 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);
 }