X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fbox-whisker.c;h=e0fbbe94b2e1d61a103520f3c96789c0ba19519c;hb=refs%2Fheads%2Frust;hp=fb5c2c62a21fefe6842f820ef6599c4e05bbf3c7;hpb=00473f96a19f9b4fc8b9dbe54dc165f2742b1140;p=pspp

diff --git a/src/math/box-whisker.c b/src/math/box-whisker.c
index fb5c2c62a2..e0fbbe94b2 100644
--- a/src/math/box-whisker.c
+++ b/src/math/box-whisker.c
@@ -22,6 +22,7 @@
 #include <float.h>
 
 #include "data/case.h"
+#include "data/data-out.h"
 #include "data/val-type.h"
 #include "data/variable.h"
 #include "libpspp/assertion.h"
@@ -39,7 +40,7 @@ destroy (struct statistic *s)
   struct order_stats *os = &bw->parent;
   struct ll *ll;
 
-  for (ll = ll_head (&bw->outliers); ll != ll_null (&bw->outliers); )
+  for (ll = ll_head (&bw->outliers); ll != ll_null (&bw->outliers);)
     {
       struct outlier *e = ll_data (ll, struct outlier, ll);
 
@@ -60,31 +61,51 @@ acc (struct statistic *s, const struct ccase *cx,
 {
   struct box_whisker *bw = UP_CAST (s, struct box_whisker, parent.parent);
   bool extreme;
-  struct outlier *o;
 
-  if ( y < bw->hinges[2] + bw->step)
-      bw->whiskers[1] = y;
+  if (y > bw->hinges[2] + bw->step) /* Upper outlier */
+    {
+      extreme = (y > bw->hinges[2] + 2 * bw->step) ;
+    }
 
-  if (bw->whiskers[0] == SYSMIS ||  bw->hinges[0] - bw->step > y)
-      bw->whiskers[0] = y;
+  else if (y < bw->hinges[0] - bw->step) /* Lower outlier */
+    {
+      extreme = (y < bw->hinges[0] - 2 * bw->step) ;
+    }
 
-  if ( y > bw->hinges[2] + bw->step)
-    extreme = (y > bw->hinges[2] + 2 * bw->step) ;
+  else /* Not an outlier */
+    {
+      if (bw->whiskers[0] == SYSMIS)
+	bw->whiskers[0] = y;
 
-  else if (y < bw->hinges[0] - bw->step)
-    extreme = (y < bw->hinges[0] - 2 * bw->step) ;
+      if (y > bw->whiskers[1])
+	bw->whiskers[1] = y;
 
-  else
-    return;
+      return;
+    }
+
+  /* y is an outlier */
 
-  o = xzalloc (sizeof *o) ;
+  struct outlier *o = XZALLOC (struct outlier);
   o->value = y;
   o->extreme = extreme;
   ds_init_empty (&o->label);
 
-  ds_put_format (&o->label,
-		   "%ld",
-                 (casenumber) case_data_idx (cx, bw->casenumber_idx)->f);
+  if (bw->id_var)
+    {
+      char *s = data_out (case_data_idx (cx, bw->id_idx),
+                          var_get_encoding (bw->id_var),
+                          var_get_print_format (bw->id_var),
+                          settings_get_fmt_settings ());
+
+      ds_put_cstr (&o->label, s);
+      free (s);
+    }
+  else
+    {
+      ds_put_format (&o->label,
+                     "%ld",
+                     (casenumber) case_num_idx (cx, bw->id_idx));
+    }
 
   ll_push_head (&bw->outliers, &o->ll);
 }
@@ -115,32 +136,32 @@ box_whisker_outliers (const struct box_whisker *bw)
 
   TH are the tukey hinges of the dataset.
 
-  Casenumber_idx is the index into the casereader which will be used to label 
+  id_idx is the index into the casereader which will be used to label
   outliers.
+  id_var is the variable from which that label came, or NULL
 */
 struct box_whisker *
 box_whisker_create (const struct tukey_hinges *th,
-		    size_t casenumber_idx)
+		    size_t id_idx, const struct variable *id_var)
 {
-  struct box_whisker *w = xzalloc (sizeof (*w));
-  struct order_stats *os = &w->parent;
-  struct statistic *stat = &os->parent;
-
-  os->n_k = 0;
-
-  stat->destroy = destroy;
-  stat->accumulate = acc;
-
-  tukey_hinges_calculate (th, w->hinges);
-
-  w->casenumber_idx = casenumber_idx;
-
-  w->step = (w->hinges[2] - w->hinges[0]) * 1.5;
-
-  w->whiskers[1] = w->hinges[2];
-  w->whiskers[0] = SYSMIS;
-
-  ll_init (&w->outliers);
+  double hinges[3];
+  tukey_hinges_calculate (th, hinges);
+
+  struct box_whisker *w = xmalloc (sizeof *w);
+  *w = (struct box_whisker) {
+    .parent = {
+      .parent = {
+        .destroy = destroy,
+      },
+      .accumulate = acc,
+    },
+    .hinges = { hinges[0], hinges[1], hinges[2] },
+    .whiskers = { SYSMIS, hinges[2] },
+    .outliers = LL_INITIALIZER (w->outliers),
+    .step = (hinges[2] - hinges[0]) * 1.5,
+    .id_idx = id_idx,
+    .id_var = id_var,
+  };
 
   return w;
 }