#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"
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);
{
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;
+ }
- o = xzalloc (sizeof *o) ;
+ /* y is an outlier */
+
+ 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);
}
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 box_whisker *w = XZALLOC (struct box_whisker);
struct order_stats *os = &w->parent;
struct statistic *stat = &os->parent;
tukey_hinges_calculate (th, w->hinges);
- w->casenumber_idx = casenumber_idx;
+ w->id_idx = id_idx;
+ w->id_var = id_var;
w->step = (w->hinges[2] - w->hinges[0]) * 1.5;