X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Ffrequencies.q;h=723d26e8341c3f7c732f3ed394067dea0b4204c6;hb=608b1765241e7c6f9bd5e86a8f81cf15edeb413b;hp=b8050792b534a1caa56198e06d0ba96e2b893582;hpb=93268947fc32952b71f386e3dd1d3d1c6152203f;p=pspp

diff --git a/src/language/stats/frequencies.q b/src/language/stats/frequencies.q
index b8050792b5..723d26e834 100644
--- a/src/language/stats/frequencies.q
+++ b/src/language/stats/frequencies.q
@@ -125,10 +125,6 @@ struct percentile
 {
   double p;        /* the %ile to be calculated */
   double value;    /* the %ile's value */
-  double x1;       /* The datum value <= the percentile */
-  double x2;       /* The datum value >= the percentile */
-  int flag;
-  int flag2;       /* Set to 1 if this percentile value has been found */
   bool show;       /* True to show this percentile in the statistics box. */
 };
 
@@ -157,9 +153,6 @@ struct frq_chart
     bool include_missing;       /* Whether to include missing values. */
   };
 
-/* Parsed command. */
-static struct cmd_frequencies cmd;
-
 /* Frequency tables. */
 
 /* Entire frequency table. */
@@ -205,9 +198,15 @@ struct frq_proc
     struct var_freqs *vars;
     size_t n_vars;
 
+    /* Percentiles to calculate and possibly display. */
     struct percentile *percentiles;
     int n_percentiles, n_show_percentiles;
 
+    /* Frequency table display. */
+    int max_categories;         /* Maximum categories to show. */
+    int sort;                   /* FRQ_AVALUE or FRQ_DVALUE
+                                   or FRQ_ACOUNT or FRQ_DCOUNT. */
+
     /* Statistics; number of statistics. */
     unsigned long stats;
     int n_stats;
@@ -216,28 +215,25 @@ struct frq_proc
     struct frq_chart *hist, *pie;
   };
 
-static void determine_charts (struct frq_proc *frq);
+static void determine_charts (struct frq_proc *,
+                              const struct cmd_frequencies *);
 
-static void calc_stats (const struct frq_proc *, const struct var_freqs *,
-                        double d[FRQ_N_STATS]);
+static void calc_stats (const struct var_freqs *, double d[FRQ_N_STATS]);
+static void calc_percentiles (const struct frq_proc *,
+                              const struct var_freqs *);
 
 static void precalc (struct frq_proc *, struct casereader *, struct dataset *);
 static void calc (struct frq_proc *, const struct ccase *,
                   const struct dataset *);
 static void postcalc (struct frq_proc *, const struct dataset *);
 
-static void postprocess_freq_tab (struct var_freqs *);
+static void postprocess_freq_tab (const struct frq_proc *, struct var_freqs *);
 static void dump_freq_table (const struct var_freqs *,
                              const struct variable *weight_var);
 static void dump_statistics (const struct frq_proc *, const struct var_freqs *,
                              const struct variable *weight_var);
 static void cleanup_freq_tab (struct var_freqs *);
 
-static algo_compare_func compare_value_numeric_a, compare_value_alpha_a;
-static algo_compare_func compare_value_numeric_d, compare_value_alpha_d;
-static algo_compare_func compare_freq_numeric_a, compare_freq_alpha_a;
-static algo_compare_func compare_freq_numeric_d, compare_freq_alpha_d;
-
 static void add_percentile (struct frq_proc *, double x, bool show,
                             size_t *allocated_percentiles);
 
@@ -253,6 +249,7 @@ struct histogram *freq_tab_to_hist(const struct frq_proc *,
 int
 cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 {
+  struct cmd_frequencies cmd;
   struct frq_proc frq;
   struct casegrouper *grouper;
   struct casereader *input, *group;
@@ -280,6 +277,12 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
       return CMD_FAILURE;
     }
 
+  /* Figure out when to show frequency tables. */
+  frq.max_categories = (cmd.table == FRQ_NOTABLE ? -1
+                        : cmd.table == FRQ_TABLE ? INT_MAX
+                        : cmd.limit);
+  frq.sort = cmd.sort;
+
   /* Figure out statistics to calculate. */
   frq.stats = 0;
   if (cmd.a_statistics[FRQ_ST_DEFAULT] || !cmd.sbc_statistics)
@@ -303,7 +306,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
       frq.n_stats++;
 
   /* Charting. */
-  determine_charts (&frq);
+  determine_charts (&frq, &cmd);
   if (cmd.sbc_histogram || cmd.sbc_piechart || cmd.sbc_ntiles)
     cmd.sort = FRQ_AVALUE;
 
@@ -373,21 +376,21 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 
 /* Figure out which charts the user requested.  */
 static void
-determine_charts (struct frq_proc *frq)
+determine_charts (struct frq_proc *frq, const struct cmd_frequencies *cmd)
 {
-  if (cmd.sbc_barchart)
+  if (cmd->sbc_barchart)
     msg (SW, _("Bar charts are not implemented."));
 
-  if (cmd.sbc_histogram)
+  if (cmd->sbc_histogram)
     {
       struct frq_chart *hist;
 
       hist = frq->hist = xmalloc (sizeof *frq->hist);
-      hist->x_min = cmd.hi_min;
-      hist->x_max = cmd.hi_max;
-      hist->y_scale = cmd.hi_scale;
-      hist->y_max = cmd.hi_scale == FRQ_FREQ ? cmd.hi_freq : cmd.hi_pcnt;
-      hist->draw_normal = cmd.hi_norm != FRQ_NONORMAL;
+      hist->x_min = cmd->hi_min;
+      hist->x_max = cmd->hi_max;
+      hist->y_scale = cmd->hi_scale;
+      hist->y_max = cmd->hi_scale == FRQ_FREQ ? cmd->hi_freq : cmd->hi_pcnt;
+      hist->draw_normal = cmd->hi_norm != FRQ_NONORMAL;
       hist->include_missing = false;
 
       if (hist->x_min != SYSMIS && hist->x_max != SYSMIS
@@ -401,15 +404,15 @@ determine_charts (struct frq_proc *frq)
         }
     }
 
-  if (cmd.sbc_piechart)
+  if (cmd->sbc_piechart)
     {
       struct frq_chart *pie;
 
       pie = frq->pie = xmalloc (sizeof *frq->pie);
-      pie->x_min = cmd.pie_min;
-      pie->x_max = cmd.pie_max;
-      pie->y_scale = cmd.pie_scale;
-      pie->include_missing = cmd.pie_missing == FRQ_MISSING;
+      pie->x_min = cmd->pie_min;
+      pie->x_max = cmd->pie_max;
+      pie->y_scale = cmd->pie_scale;
+      pie->include_missing = cmd->pie_missing == FRQ_MISSING;
 
       if (pie->x_min != SYSMIS && pie->x_max != SYSMIS
           && pie->x_min >= pie->x_max)
@@ -475,14 +478,11 @@ postcalc (struct frq_proc *frq, const struct dataset *ds)
   for (i = 0; i < frq->n_vars; i++)
     {
       struct var_freqs *vf = &frq->vars[i];
-      int n_categories;
 
-      postprocess_freq_tab (vf);
+      postprocess_freq_tab (frq, vf);
 
       /* Frequencies tables. */
-      n_categories = vf->tab.n_valid + vf->tab.n_missing;
-      if  (cmd.table == FRQ_TABLE
-           || (cmd.table == FRQ_LIMIT && n_categories <= cmd.limit))
+      if (vf->tab.n_valid + vf->tab.n_missing <= frq->max_categories)
         dump_freq_table (vf, wv);
 
       /* Statistics. */
@@ -494,7 +494,7 @@ postcalc (struct frq_proc *frq, const struct dataset *ds)
 	  double d[FRQ_N_STATS];
 	  struct histogram *histogram;
 
-	  calc_stats (frq, vf, d);
+	  calc_stats (vf, d);
 
 	  histogram = freq_tab_to_hist (frq, &vf->tab, vf->var);
 
@@ -516,28 +516,6 @@ postcalc (struct frq_proc *frq, const struct dataset *ds)
     }
 }
 
-/* Returns the comparison function that should be used for
-   sorting a frequency table by FRQ_SORT using VAL_TYPE
-   values. */
-static algo_compare_func *
-get_freq_comparator (int frq_sort, enum val_type val_type)
-{
-  bool is_numeric = val_type == VAL_NUMERIC;
-  switch (frq_sort)
-    {
-    case FRQ_AVALUE:
-      return is_numeric ? compare_value_numeric_a : compare_value_alpha_a;
-    case FRQ_DVALUE:
-      return is_numeric ? compare_value_numeric_d : compare_value_alpha_d;
-    case FRQ_AFREQ:
-      return is_numeric ? compare_freq_numeric_a : compare_freq_alpha_a;
-    case FRQ_DFREQ:
-      return is_numeric ? compare_freq_numeric_d : compare_freq_alpha_d;
-    default:
-      NOT_REACHED ();
-    }
-}
-
 /* Returns true iff the value in struct freq F is non-missing
    for variable V. */
 static bool
@@ -549,12 +527,39 @@ not_missing (const void *f_, const void *v_)
   return !var_is_value_missing (v, &f->value, MV_ANY);
 }
 
+struct freq_compare_aux
+  {
+    bool by_freq;
+    bool ascending_freq;
+
+    int width;
+    bool ascending_value;
+  };
+
+static int
+compare_freq (const void *a_, const void *b_, const void *aux_)
+{
+  const struct freq_compare_aux *aux = aux_;
+  const struct freq *a = a_;
+  const struct freq *b = b_;
+
+  if (aux->by_freq && a->count != b->count)
+    {
+      int cmp = a->count > b->count ? 1 : -1;
+      return aux->ascending_freq ? cmp : -cmp;
+    }
+  else
+    {
+      int cmp = value_compare_3way (&a->value, &b->value, aux->width);
+      return aux->ascending_value ? cmp : -cmp;
+    }
+}
 /* Summarizes the frequency table data for variable V. */
 static void
-postprocess_freq_tab (struct var_freqs *vf)
+postprocess_freq_tab (const struct frq_proc *frq, struct var_freqs *vf)
 {
   struct freq_tab *ft = &vf->tab;
-  algo_compare_func *compare;
+  struct freq_compare_aux aux;
   size_t count;
   struct freq *freqs, *f;
   size_t i;
@@ -570,9 +575,12 @@ postprocess_freq_tab (struct var_freqs *vf)
   ft->n_missing = count - ft->n_valid;
 
   /* Sort data. */
-  compare = get_freq_comparator (cmd.sort, var_get_type (vf->var));
-  sort (ft->valid, ft->n_valid, sizeof *ft->valid, compare, vf);
-  sort (ft->missing, ft->n_missing, sizeof *ft->missing, compare, vf);
+  aux.by_freq = frq->sort == FRQ_AFREQ || frq->sort == FRQ_DFREQ;
+  aux.ascending_freq = frq->sort != FRQ_DFREQ;
+  aux.width = vf->width;
+  aux.ascending_value = frq->sort != FRQ_DVALUE;
+  sort (ft->valid, ft->n_valid, sizeof *ft->valid, compare_freq, &aux);
+  sort (ft->missing, ft->n_missing, sizeof *ft->missing, compare_freq, &aux);
 
   /* Summary statistics. */
   ft->valid_cases = 0.0;
@@ -784,122 +792,6 @@ add_percentile (struct frq_proc *frq, double x, bool show,
 
 /* Comparison functions. */
 
-/* Ascending numeric compare of values. */
-static int
-compare_value_numeric_a (const void *a_, const void *b_,
-                         const void *vf_ UNUSED)
-{
-  const struct freq *a = a_;
-  const struct freq *b = b_;
-
-  if (a->value.f > b->value.f)
-    return 1;
-  else if (a->value.f < b->value.f)
-    return -1;
-  else
-    return 0;
-}
-
-/* Ascending string compare of values. */
-static int
-compare_value_alpha_a (const void *a_, const void *b_, const void *vf_)
-{
-  const struct freq *a = a_;
-  const struct freq *b = b_;
-  const struct var_freqs *vf = vf_;
-
-  return value_compare_3way (&a->value, &b->value, vf->width);
-}
-
-/* Descending numeric compare of values. */
-static int
-compare_value_numeric_d (const void *a, const void *b, const void *vf_ UNUSED)
-{
-  return -compare_value_numeric_a (a, b, vf_);
-}
-
-/* Descending string compare of values. */
-static int
-compare_value_alpha_d (const void *a, const void *b, const void *vf_)
-{
-  return -compare_value_alpha_a (a, b, vf_);
-}
-
-/* Ascending numeric compare of frequency;
-   secondary key on ascending numeric value. */
-static int
-compare_freq_numeric_a (const void *a_, const void *b_, const void *vf_ UNUSED)
-{
-  const struct freq *a = a_;
-  const struct freq *b = b_;
-
-  if (a->count > b->count)
-    return 1;
-  else if (a->count < b->count)
-    return -1;
-
-  if (a->value.f > b->value.f)
-    return 1;
-  else if (a->value.f < b->value.f)
-    return -1;
-  else
-    return 0;
-}
-
-/* Ascending numeric compare of frequency;
-   secondary key on ascending string value. */
-static int
-compare_freq_alpha_a (const void *a_, const void *b_, const void *vf_)
-{
-  const struct freq *a = a_;
-  const struct freq *b = b_;
-  const struct var_freqs *vf = vf_;
-
-  if (a->count > b->count)
-    return 1;
-  else if (a->count < b->count)
-    return -1;
-  else
-    return value_compare_3way (&a->value, &b->value, vf->width);
-}
-
-/* Descending numeric compare of frequency;
-   secondary key on ascending numeric value. */
-static int
-compare_freq_numeric_d (const void *a_, const void *b_, const void *vf_ UNUSED)
-{
-  const struct freq *a = a_;
-  const struct freq *b = b_;
-
-  if (a->count > b->count)
-    return -1;
-  else if (a->count < b->count)
-    return 1;
-
-  if (a->value.f > b->value.f)
-    return 1;
-  else if (a->value.f < b->value.f)
-    return -1;
-  else
-    return 0;
-}
-
-/* Descending numeric compare of frequency;
-   secondary key on ascending string value. */
-static int
-compare_freq_alpha_d (const void *a_, const void *b_, const void *vf_)
-{
-  const struct freq *a = a_;
-  const struct freq *b = b_;
-  const struct var_freqs *vf = vf_;
-
-  if (a->count > b->count)
-    return -1;
-  else if (a->count < b->count)
-    return 1;
-  else
-    return value_compare_3way (&a->value, &b->value, vf->width);
-}
 
 /* Frequency table display. */
 
@@ -988,114 +880,74 @@ dump_freq_table (const struct var_freqs *vf, const struct variable *wv)
 
 /* Statistical display. */
 
-/* Calculates all the pertinent statistics for variable V, putting them in
-   array D[]. */
+static double
+calc_percentile (double p, double valid_cases, double x1, double x2)
+{
+  double s, dummy;
+
+  s = (settings_get_algorithm () != COMPATIBLE
+       ? modf ((valid_cases - 1) * p, &dummy)
+       : modf ((valid_cases + 1) * p - 1, &dummy));
+
+  return x1 + (x2 - x1) * s;
+}
+
+/* Calculates all of the percentiles for VF within FRQ. */
 static void
-calc_stats (const struct frq_proc *frq,
-            const struct var_freqs *vf, double d[FRQ_N_STATS])
+calc_percentiles (const struct frq_proc *frq, const struct var_freqs *vf)
 {
   const struct freq_tab *ft = &vf->tab;
   double W = ft->valid_cases;
-  struct moments *m;
-  struct freq *f=0;
-  double prev_value;
-  int most_often;
-  double X_mode;
-
+  const struct freq *f;
+  int percentile_idx;
   double rank;
-  int i = 0;
-  int idx;
-
-  /* Calculate percentiles. */
 
   assert (ft->n_valid > 0);
 
-  for (i = 0; i < frq->n_percentiles; i++)
-    {
-      struct percentile *pc = &frq->percentiles[i];
-
-      pc->flag = 0;
-      pc->flag2 = 0;
-    }
-
   rank = 0;
-  prev_value = SYSMIS;
-  for (idx = 0; idx < ft->n_valid; ++idx)
+  percentile_idx = 0;
+  for (f = ft->valid; f < ft->missing; f++)
     {
-      f = &ft->valid[idx];
-      rank += f->count ;
-      for (i = 0; i < frq->n_percentiles; i++)
+      rank += f->count;
+      for (; percentile_idx < frq->n_percentiles; percentile_idx++)
         {
-          struct percentile *pc = &frq->percentiles[i];
-	  double tp;
-
-	  if ( pc->flag2  ) continue ;
-
-	  if ( settings_get_algorithm () != COMPATIBLE )
-	    tp =
-	      (ft->valid_cases - 1) *  pc->p;
-	  else
-	    tp =
-	      (ft->valid_cases + 1) *  pc->p - 1;
-
-	  if ( pc->flag )
-	    {
-	      pc->x2 = f->value.f;
-	      pc->x1 = prev_value;
-	      pc->flag2 = 1;
-	      continue;
-	    }
-
-          if (rank >  tp )
-	  {
-	    if ( f->count > 1 && rank - (f->count - 1) > tp )
-	      {
-		pc->x2 = pc->x1 = f->value.f;
-		pc->flag2 = 1;
-	      }
-	    else
-	      {
-		pc->flag=1;
-	      }
+          struct percentile *pc = &frq->percentiles[percentile_idx];
+          double tp;
 
-	    continue;
-	  }
-        }
-      prev_value = f->value.f;
-    }
+          tp = (settings_get_algorithm () == ENHANCED
+                ? (W - 1) * pc->p
+                : (W + 1) * pc->p - 1);
 
-  for (i = 0; i < frq->n_percentiles; i++)
-    {
-      struct percentile *pc = &frq->percentiles[i];
-
-      /* Catches the case when p == 100% */
-      if ( ! pc->flag2 )
-	pc->x1 = pc->x2 = f->value.f;
+          if (rank <= tp)
+            break;
 
-      /*
-      printf("percentile %d (p==%.2f); X1 = %g; X2 = %g\n",
-	     i,pc->p,pc->x1,pc->x2);
-      */
+          if (f->count > 1
+              && (rank - (f->count - 1) > tp || f + 1 >= ft->missing))
+            pc->value = f->value.f;
+          else
+            pc->value = calc_percentile (pc->p, W, f->value.f, f[1].value.f);
+        }
     }
-
-  for (i = 0; i < frq->n_percentiles; i++)
+  for (; percentile_idx < frq->n_percentiles; percentile_idx++)
     {
-      struct percentile *pc = &frq->percentiles[i];
-      double s;
-
-      double dummy;
-      if ( settings_get_algorithm () != COMPATIBLE )
-	{
-	  s = modf((ft->valid_cases - 1) * pc->p , &dummy);
-	}
-      else
-	{
-	  s = modf((ft->valid_cases + 1) * pc->p -1, &dummy);
-	}
-
-      pc->value = pc->x1 + (pc->x2 - pc->x1) * s ;
+      struct percentile *pc = &frq->percentiles[percentile_idx];
+      pc->value = ft->valid[ft->n_valid - 1].value.f;
     }
+}
 
+/* Calculates all the pertinent statistics for VF, putting them in array
+   D[]. */
+static void
+calc_stats (const struct var_freqs *vf, double d[FRQ_N_STATS])
+{
+  const struct freq_tab *ft = &vf->tab;
+  double W = ft->valid_cases;
+  const struct freq *f;
+  struct moments *m;
+  int most_often;
+  double X_mode;
+
+  assert (ft->n_valid > 0);
 
   /* Calculate the mode. */
   most_often = -1;
@@ -1157,7 +1009,8 @@ dump_statistics (const struct frq_proc *frq, const struct var_freqs *vf,
 	   var_get_name (vf->var));
       return;
     }
-  calc_stats (frq, vf, stat_value);
+  calc_stats (vf, stat_value);
+  calc_percentiles (frq, vf);
 
   t = tab_create (3, frq->n_stats + frq->n_show_percentiles + 2);