X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Froc.c;h=0709992ce469ae113814f6996b549fe0d684b7fc;hb=1fe8ef2486ef1e853d68bbab967ccb812900128f;hp=6a0e412edcadd40be8afd9c6044eea4ce20fc5ce;hpb=d36d05a067ec2653a52e54bd61d21d8067c47be2;p=pspp-builds.git

diff --git a/src/language/stats/roc.c b/src/language/stats/roc.c
index 6a0e412e..0709992c 100644
--- 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);
 
@@ -540,77 +564,92 @@ append_cutpoint (struct casewriter *writer, double cutpoint)
 }
 
 
+/* Prepare the cutpoints */
 static void
-do_roc (struct cmd_roc *roc, struct casereader *input, struct dictionary *dict)
+prepare_cutpoints (struct cmd_roc *roc, struct roc_state *rs, struct casereader *input)
 {
   int i;
+  struct casereader *r = casereader_clone (input);
+  struct ccase *c;
+  struct caseproto *proto = caseproto_create ();
 
-  struct roc_state *rs = xcalloc (roc->n_vars, sizeof *rs);
+  struct subcase ordering;
+  subcase_init (&ordering, CUTPOINT, 0, SC_ASCEND);
 
-  struct casewriter *neg_wtr = autopaging_writer_create (casereader_get_proto (input));
+  proto = caseproto_add_width (proto, 0); /* cutpoint */
+  proto = caseproto_add_width (proto, 0); /* TP */
+  proto = caseproto_add_width (proto, 0); /* FN */
+  proto = caseproto_add_width (proto, 0); /* TN */
+  proto = caseproto_add_width (proto, 0); /* FP */
 
-  struct casereader *negatives = NULL;
-  struct casereader *positives = NULL;
+  for (i = 0 ; i < roc->n_vars; ++i)
+    {
+      rs[i].cutpoint_wtr = sort_create_writer (&ordering, proto);
+      rs[i].prev_result = SYSMIS;
+      rs[i].max = -DBL_MAX;
+      rs[i].min = DBL_MAX;
+    }
 
-  struct caseproto *n_proto = caseproto_create ();
+  for (; (c = casereader_read (r)) != NULL; case_unref (c))
+    {
+      for (i = 0 ; i < roc->n_vars; ++i)
+	{
+	  const union value *v = case_data (c, roc->vars[i]); 
+	  const double result = v->f;
 
-  struct subcase up_ordering;
-  struct subcase down_ordering;
+	  if ( mv_is_value_missing (var_get_missing_values (roc->vars[i]), v, roc->exclude))
+	    continue;
 
-  /* Prepare the cutpoints */
-  {
-    struct casereader *r = casereader_clone (input);
-    struct ccase *c;
-    struct caseproto *proto = caseproto_create ();
+	  minimize (&rs[i].min, result);
+	  maximize (&rs[i].max, result);
 
-    struct subcase ordering;
-    subcase_init (&ordering, CUTPOINT, 0, SC_ASCEND);
+	  if ( rs[i].prev_result != SYSMIS && rs[i].prev_result != result )
+	    {
+	      const double mean = (result + rs[i].prev_result ) / 2.0;
+	      append_cutpoint (rs[i].cutpoint_wtr, mean);
+	    }
 
-    proto = caseproto_add_width (proto, 0); /* cutpoint */
-    proto = caseproto_add_width (proto, 0); /* TP */
-    proto = caseproto_add_width (proto, 0); /* FN */
-    proto = caseproto_add_width (proto, 0); /* TN */
-    proto = caseproto_add_width (proto, 0); /* FP */
+	  rs[i].prev_result = result;
+	}
+    }
+  casereader_destroy (r);
 
 
-    for (i = 0 ; i < roc->n_vars; ++i)
-      {
-	rs[i].cutpoint_wtr = sort_create_writer (&ordering, proto);
-	rs[i].prev_result = SYSMIS;
-	rs[i].max = -DBL_MAX;
-	rs[i].min = DBL_MAX;
-      }
+  /* Append the min and max cutpoints */
+  for (i = 0 ; i < roc->n_vars; ++i)
+    {
+      append_cutpoint (rs[i].cutpoint_wtr, rs[i].min - 1);
+      append_cutpoint (rs[i].cutpoint_wtr, rs[i].max + 1);
 
-    for (; (c = casereader_read (r)) != NULL; case_unref (c))
-      {
-	for (i = 0 ; i < roc->n_vars; ++i)
-	  {
-	    const double result = case_data (c, roc->vars[i])->f;
+      rs[i].cutpoint_rdr = casewriter_make_reader (rs[i].cutpoint_wtr);
+    }
+}
 
-	    minimize (&rs[i].min, result);
-	    maximize (&rs[i].max, result);
+static void
+do_roc (struct cmd_roc *roc, struct casereader *reader, struct dictionary *dict)
+{
+  int i;
 
-	    if ( rs[i].prev_result != SYSMIS && rs[i].prev_result != result )
-	      {
-		const double mean = (result + rs[i].prev_result ) / 2.0;
-		append_cutpoint (rs[i].cutpoint_wtr, mean);
-	      }
+  struct roc_state *rs = xcalloc (roc->n_vars, sizeof *rs);
 
-	    rs[i].prev_result = result;
-	  }
-      }
-    casereader_destroy (r);
+  struct casereader *negatives = NULL;
+  struct casereader *positives = NULL;
 
+  struct caseproto *n_proto = caseproto_create ();
 
-    /* Append the min and max cutpoints */
-    for (i = 0 ; i < roc->n_vars; ++i)
-      {
-	append_cutpoint (rs[i].cutpoint_wtr, rs[i].min - 1);
-	append_cutpoint (rs[i].cutpoint_wtr, rs[i].max + 1);
+  struct subcase up_ordering;
+  struct subcase down_ordering;
 
-	rs[i].cutpoint_rdr = casewriter_make_reader (rs[i].cutpoint_wtr);
-      }
-  }
+  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 = 
     casereader_create_filter_func (input,
@@ -944,13 +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);
 }