X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Ffrequencies.c;h=307a869ba2d1e17d723c80443f19b8ba7a7a7240;hb=5e755f453ddba1a9f0c9677d9484950d5c21f4fd;hp=8bd6c600205d3fe75dc6a99dd230c2ab9591146b;hpb=5cab4cf3322f29c0ed7134d23740e07382914f20;p=pspp

diff --git a/src/language/stats/frequencies.c b/src/language/stats/frequencies.c
index 8bd6c60020..307a869ba2 100644
--- a/src/language/stats/frequencies.c
+++ b/src/language/stats/frequencies.c
@@ -45,14 +45,12 @@
 #include "libpspp/hmap.h"
 #include "libpspp/message.h"
 #include "libpspp/misc.h"
-#include "libpspp/pool.h"
 
 #include "math/histogram.h"
 #include "math/moments.h"
 #include "math/chart-geometry.h"
 
 
-#include "output/chart-item.h"
 #include "output/charts/barchart.h"
 #include "output/charts/piechart.h"
 #include "output/charts/plot-hist.h"
@@ -70,7 +68,6 @@
 struct percentile
 {
   double p;        /* the %ile to be calculated */
-  double value;    /* the %ile's value */
   bool show;       /* True to show this percentile in the statistics box. */
 };
 
@@ -191,12 +188,9 @@ struct var_freqs
     /* Freqency table. */
     struct freq_tab tab;	/* Frequencies table to use. */
 
-    /* Percentiles. */
-    int n_groups;		/* Number of groups. */
-    double *groups;		/* Groups. */
-
     /* Statistics. */
     double stat[FRQ_ST_count];
+    double *percentiles;
 
     /* Variable attributes. */
     int width;
@@ -204,15 +198,13 @@ struct var_freqs
 
 struct frq_proc
   {
-    struct pool *pool;
-
     struct var_freqs *vars;
     size_t n_vars;
 
     /* Percentiles to calculate and possibly display. */
     struct percentile *percentiles;
-    const struct percentile *median;
-    int n_percentiles;
+    size_t median_idx;
+    size_t n_percentiles;
 
     /* Frequency table display. */
     long int max_categories;         /* Maximum categories to show. */
@@ -225,6 +217,8 @@ struct frq_proc
 
     /* Histogram and pie chart settings. */
     struct frq_chart *hist, *pie, *bar;
+
+    bool warn;
   };
 
 
@@ -271,9 +265,8 @@ compare_freq (const void *a_, const void *b_, const void *aux_)
 }
 
 /* Create a gsl_histogram from a freq_tab */
-static struct histogram *
-freq_tab_to_hist (const struct frq_proc *frq, const struct freq_tab *ft,
-                  const struct variable *var);
+static struct histogram *freq_tab_to_hist (const struct frq_proc *,
+                                           const struct var_freqs *);
 
 static void
 put_freq_row (struct pivot_table *table, int var_idx,
@@ -294,7 +287,7 @@ dump_freq_table (const struct var_freqs *vf, const struct variable *wv)
   const struct freq_tab *ft = &vf->tab;
 
   struct pivot_table *table = pivot_table_create__ (pivot_value_new_variable (
-                                                      vf->var));
+                                                      vf->var), "Frequencies");
   pivot_table_set_weight_var (table, wv);
 
   pivot_dimension_create (table, PIVOT_AXIS_COLUMN, N_("Statistics"),
@@ -361,20 +354,24 @@ calc_percentile (double p, double valid_cases, double x1, double x2)
 
 /* Calculates all of the percentiles for VF within FRQ. */
 static void
-calc_percentiles (const struct frq_proc *frq, const struct var_freqs *vf)
+calc_percentiles (const struct frq_proc *frq, struct var_freqs *vf)
 {
+  if (!frq->n_percentiles)
+    return;
+
+  vf->percentiles = xnmalloc (frq->n_percentiles, sizeof *vf->percentiles);
+
   const struct freq_tab *ft = &vf->tab;
-  double W = ft->valid_cases;
-  const struct freq *f;
-  int percentile_idx = 0;
-  double  rank = 0;
+  const double W = ft->valid_cases;
+  size_t idx = 0;
 
-  for (f = ft->valid; f < ft->missing; f++)
+  double rank = 0;
+  for (const struct freq *f = ft->valid; f < ft->missing; f++)
     {
       rank += f->count;
-      for (; percentile_idx < frq->n_percentiles; percentile_idx++)
+      for (; idx < frq->n_percentiles; idx++)
         {
-          struct percentile *pc = &frq->percentiles[percentile_idx];
+          struct percentile *pc = &frq->percentiles[idx];
           double tp;
 
           tp = (settings_get_algorithm () == ENHANCED
@@ -385,18 +382,16 @@ calc_percentiles (const struct frq_proc *frq, const struct var_freqs *vf)
             break;
 
           if (tp + 1 < rank || f + 1 >= ft->missing)
-            pc->value = f->values[0].f;
+            vf->percentiles[idx] = f->values[0].f;
           else
-            pc->value = calc_percentile (pc->p, W, f->values[0].f, f[1].values[0].f);
+            vf->percentiles[idx] = calc_percentile (pc->p, W, f->values[0].f,
+                                                    f[1].values[0].f);
         }
     }
-  for (; percentile_idx < frq->n_percentiles; percentile_idx++)
-    {
-      struct percentile *pc = &frq->percentiles[percentile_idx];
-      pc->value = (ft->n_valid > 0
-                   ? ft->valid[ft->n_valid - 1].values[0].f
-                   : SYSMIS);
-    }
+  for (; idx < frq->n_percentiles; idx++)
+    vf->percentiles[idx] = (ft->n_valid > 0
+                            ? ft->valid[ft->n_valid - 1].values[0].f
+                            : SYSMIS);
 }
 
 /* Returns true iff the value in struct freq F is non-missing
@@ -407,7 +402,7 @@ not_missing (const void *f_, const void *v_)
   const struct freq *f = f_;
   const struct variable *v = v_;
 
-  return !var_is_value_missing (v, f->values, MV_ANY);
+  return !var_is_value_missing (v, f->values);
 }
 
 
@@ -441,7 +436,7 @@ postprocess_freq_tab (const struct frq_proc *frq, struct var_freqs *vf)
 
   /* Summary statistics. */
   ft->valid_cases = 0.0;
-  for(i = 0 ;  i < ft->n_valid ; ++i )
+  for(i = 0 ;  i < ft->n_valid ; ++i)
     {
       f = &ft->valid[i];
       ft->valid_cases += f->count;
@@ -449,7 +444,7 @@ postprocess_freq_tab (const struct frq_proc *frq, struct var_freqs *vf)
     }
 
   ft->total_cases = ft->valid_cases ;
-  for(i = 0 ;  i < ft->n_missing ; ++i )
+  for(i = 0 ;  i < ft->n_missing ; ++i)
     {
       f = &ft->missing[i];
       ft->total_cases += f->count;
@@ -469,7 +464,7 @@ cleanup_freq_tab (struct var_freqs *vf)
 static void
 calc (struct frq_proc *frq, const struct ccase *c, const struct dataset *ds)
 {
-  double weight = dict_get_case_weight (dataset_dict (ds), c, NULL);
+  double weight = dict_get_case_weight (dataset_dict (ds), c, &frq->warn);
   size_t i;
 
   for (i = 0; i < frq->n_vars; i++)
@@ -541,16 +536,16 @@ postcalc (struct frq_proc *frq, const struct dataset *ds)
 
 	  calc_stats (frq, vf, d);
 
-	  histogram = freq_tab_to_hist (frq, &vf->tab, vf->var);
+	  histogram = freq_tab_to_hist (frq, vf);
 
-	  if ( histogram)
+	  if (histogram)
 	    {
-	      chart_item_submit (histogram_chart_create (
-                               histogram->gsl_hist, var_to_string(vf->var),
-                               vf->tab.valid_cases,
-                               d[FRQ_ST_MEAN],
-                               d[FRQ_ST_STDDEV],
-                               frq->hist->draw_normal));
+	      chart_submit (histogram_chart_create (
+                              histogram->gsl_hist, var_to_string(vf->var),
+                              vf->tab.valid_cases,
+                              d[FRQ_ST_MEAN],
+                              d[FRQ_ST_STDDEV],
+                              frq->hist->draw_normal));
 
 	      statistic_destroy (&histogram->parent);
 	    }
@@ -592,7 +587,6 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
   int hi_pcnt = INT_MIN;
   int hi_norm = FRQ_NONORMAL;
 
-  frq.pool = pool_create ();
   frq.sort = FRQ_AVALUE;
 
   frq.vars = NULL;
@@ -613,13 +607,14 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
   frq.hist = NULL;
   frq.pie = NULL;
   frq.bar = NULL;
+  frq.warn = true;
 
 
   /* Accept an optional, completely pointless "/VARIABLES=" */
   lex_match (lexer, T_SLASH);
   if (lex_match_id  (lexer, "VARIABLES"))
     {
-      if (! lex_force_match (lexer, T_EQUALS) )
+      if (! lex_force_match (lexer, T_EQUALS))
         goto error;
     }
 
@@ -629,7 +624,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 			      PV_NO_DUPLICATE))
     goto error;
 
-  frq.vars = xzalloc (frq.n_vars * sizeof (*frq.vars));
+  frq.vars = xcalloc (frq.n_vars, sizeof (*frq.vars));
   for (i = 0; i < frq.n_vars; ++i)
     {
       frq.vars[i].var = vars[i];
@@ -795,7 +790,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
               else if (lex_match_id (lexer, "LIMIT"))
                 {
                   if (!lex_force_match (lexer, T_LPAREN)
-                      || !lex_force_int (lexer))
+                      || !lex_force_int_range (lexer, "LIMIT", 0, INT_MAX))
                     goto error;
 
                   frq.max_categories = lex_integer (lexer);
@@ -831,12 +826,11 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
         {
 	  lex_match (lexer, T_EQUALS);
 
-	  if (lex_force_int (lexer))
+	  if (lex_force_int_range (lexer, "NTILES", 0, INT_MAX))
 	    {
-	      int i;
 	      int n = lex_integer (lexer);
 	      lex_get (lexer);
-	      for (i = 0; i < n + 1; ++i)
+	      for (int i = 0; i < n + 1; ++i)
 		{
 		  frq.percentiles =
 		    xrealloc (frq.percentiles,
@@ -894,13 +888,9 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
                   hi_scale = FRQ_FREQ;
                   if (lex_match (lexer, T_LPAREN))
                     {
-                      if (lex_force_int (lexer))
+                      if (lex_force_int_range (lexer, "FREQ", 1, INT_MAX))
                         {
 			  hi_freq = lex_integer (lexer);
-			  if (hi_freq <= 0)
-			    {
-			      lex_error (lexer, _("Histogram frequency must be greater than zero."));
-			    }
 			  lex_get (lexer);
 			  if (! lex_force_match (lexer, T_RPAREN))
 			    goto error;
@@ -912,13 +902,9 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
                   hi_scale = FRQ_PERCENT;
                   if (lex_match (lexer, T_LPAREN))
                     {
-                      if (lex_force_int (lexer))
+                      if (lex_force_int_range (lexer, "PERCENT", 1, INT_MAX))
 			{
 			  hi_pcnt = lex_integer (lexer);
-			  if (hi_pcnt <= 0)
-			    {
-			      lex_error (lexer, _("Histogram percentage must be greater than zero."));
-			    }
 			  lex_get (lexer);
 			  if (! lex_force_match (lexer, T_RPAREN))
 			    goto error;
@@ -1034,7 +1020,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 		}
 	      else if (lex_match_id (lexer, "FREQ"))
 		{
-		  if ( lex_match (lexer, T_LPAREN))
+		  if (lex_match (lexer, T_LPAREN))
 		    {
 		      if (lex_force_num (lexer))
 			{
@@ -1048,7 +1034,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 		}
 	      else if (lex_match_id (lexer, "PERCENT"))
 		{
-		  if ( lex_match (lexer, T_LPAREN))
+		  if (lex_match (lexer, T_LPAREN))
 		    {
 		      if (lex_force_num (lexer))
 			{
@@ -1215,11 +1201,11 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 
     frq.n_percentiles = o;
 
-    frq.median = NULL;
+    frq.median_idx = SIZE_MAX;
     for (i = 0; i < frq.n_percentiles; i++)
       if (frq.percentiles[i].p == 0.5)
         {
-          frq.median = &frq.percentiles[i];
+          frq.median_idx = i;
           break;
         }
   }
@@ -1244,14 +1230,14 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
     ok = proc_commit (ds) && ok;
   }
 
-
   free (vars);
+  for (size_t i = 0; i < frq.n_vars; i++)
+    free (frq.vars[i].percentiles);
   free (frq.vars);
   free (frq.bar);
   free (frq.pie);
   free (frq.hist);
   free (frq.percentiles);
-  pool_destroy (frq.pool);
 
   return CMD_SUCCESS;
 
@@ -1259,17 +1245,18 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 
   free (vars);
   free (frq.vars);
+  for (size_t i = 0; i < frq.n_vars; i++)
+    free (frq.vars[i].percentiles);
   free (frq.bar);
   free (frq.pie);
   free (frq.hist);
   free (frq.percentiles);
-  pool_destroy (frq.pool);
 
   return CMD_FAILURE;
 }
 
 static double
-calculate_iqr (const struct frq_proc *frq)
+calculate_iqr (const struct frq_proc *frq, const struct var_freqs *vf)
 {
   double q1 = SYSMIS;
   double q3 = SYSMIS;
@@ -1282,9 +1269,9 @@ calculate_iqr (const struct frq_proc *frq)
       struct percentile *pc = &frq->percentiles[i];
 
       if (fabs (0.25 - pc->p) < DBL_EPSILON)
-        q1 = pc->value;
+        q1 = vf->percentiles[i];
       else if (fabs (0.75 - pc->p) < DBL_EPSILON)
-        q3 = pc->value;
+        q3 = vf->percentiles[i];
     }
 
   return q1 == SYSMIS || q3 == SYSMIS ? SYSMIS : q3 - q1;
@@ -1295,7 +1282,7 @@ chart_includes_value (const struct frq_chart *chart,
                       const struct variable *var,
                       const union value *value)
 {
-  if (!chart->include_missing && var_is_value_missing (var, value, MV_ANY))
+  if (!chart->include_missing && var_is_value_missing (var, value))
     return false;
 
   if (var_is_numeric (var)
@@ -1308,24 +1295,17 @@ chart_includes_value (const struct frq_chart *chart,
 
 /* Create a gsl_histogram from a freq_tab */
 static struct histogram *
-freq_tab_to_hist (const struct frq_proc *frq, const struct freq_tab *ft,
-                  const struct variable *var)
+freq_tab_to_hist (const struct frq_proc *frq, const struct var_freqs *vf)
 {
-  double x_min, x_max, valid_freq;
-  int i;
-  double bin_width;
-  struct histogram *histogram;
-  double iqr;
-
   /* Find out the extremes of the x value, within the range to be included in
      the histogram, and sum the total frequency of those values. */
-  x_min = DBL_MAX;
-  x_max = -DBL_MAX;
-  valid_freq = 0;
-  for (i = 0; i < ft->n_valid; i++)
+  double x_min = DBL_MAX;
+  double x_max = -DBL_MAX;
+  double valid_freq = 0;
+  for (int i = 0; i < vf->tab.n_valid; i++)
     {
-      const struct freq *f = &ft->valid[i];
-      if (chart_includes_value (frq->hist, var, f->values))
+      const struct freq *f = &vf->tab.valid[i];
+      if (chart_includes_value (frq->hist, vf->var, f->values))
         {
           x_min = MIN (x_min, f->values[0].f);
           x_max = MAX (x_max, f->values[0].f);
@@ -1336,25 +1316,21 @@ freq_tab_to_hist (const struct frq_proc *frq, const struct freq_tab *ft,
   if (valid_freq <= 0)
     return NULL;
 
-  iqr = calculate_iqr (frq);
-
-  if (iqr > 0)
-    /* Freedman-Diaconis' choice of bin width. */
-    bin_width = 2 * iqr / pow (valid_freq, 1.0 / 3.0);
+  double iqr = calculate_iqr (frq, vf);
 
-  else
-    /* Sturges Rule */
-    bin_width = (x_max - x_min) / (1 + log2 (valid_freq));
+  double bin_width =
+    (iqr > 0
+     ? 2 * iqr / pow (valid_freq, 1.0 / 3.0)       /* Freedman-Diaconis. */
+     : (x_max - x_min) / (1 + log2 (valid_freq))); /* Sturges */
 
-  histogram = histogram_create (bin_width, x_min, x_max);
-
-  if ( histogram == NULL)
+  struct histogram *histogram = histogram_create (bin_width, x_min, x_max);
+  if (histogram == NULL)
     return NULL;
 
-  for (i = 0; i < ft->n_valid; i++)
+  for (int i = 0; i < vf->tab.n_valid; i++)
     {
-      const struct freq *f = &ft->valid[i];
-      if (chart_includes_value (frq->hist, var, f->values))
+      const struct freq *f = &vf->tab.valid[i];
+      if (chart_includes_value (frq->hist, vf->var, f->values))
         histogram_add (histogram, f->values[0].f, f->count);
     }
 
@@ -1474,7 +1450,7 @@ do_piechart(const struct frq_chart *pie, const struct variable *var,
     msg (SW, _("Omitting pie chart for %s, which has over 50 unique values."),
          var_get_name (var));
   else
-    chart_item_submit (piechart_create (var, slices, n_slices));
+    chart_submit (piechart_create (var, slices, n_slices));
 
   free (slices);
 }
@@ -1487,10 +1463,13 @@ do_barchart(const struct frq_chart *bar, const struct variable **var,
   int n_slices;
   struct freq **slices = pick_cat_counts_ptr (bar, frq_tab, &n_slices);
 
-  chart_item_submit (barchart_create (var, 1,
-				      (bar->y_scale == FRQ_FREQ) ? _("Count") : _("Percent"),
-				      (bar->y_scale == FRQ_PERCENT),
-				      slices, n_slices));
+  if (n_slices < 1)
+    msg (SW, _("Omitting bar chart, which has no values."));
+  else
+    chart_submit (barchart_create (var, 1,
+                                   (bar->y_scale == FRQ_FREQ) ? _("Count") : _("Percent"),
+                                   (bar->y_scale == FRQ_PERCENT),
+                                   slices, n_slices));
   free (slices);
 }
 
@@ -1553,7 +1532,9 @@ calc_stats (const struct frq_proc *frq, const struct var_freqs *vf,
   d[FRQ_ST_SEMEAN] = d[FRQ_ST_STDDEV] / sqrt (W);
   d[FRQ_ST_SESKEWNESS] = calc_seskew (W);
   d[FRQ_ST_SEKURTOSIS] = calc_sekurt (W);
-  d[FRQ_ST_MEDIAN] = frq->median ? frq->median->value : SYSMIS;
+  d[FRQ_ST_MEDIAN] = (frq->median_idx != SIZE_MAX
+                      ? vf->percentiles[frq->median_idx]
+                      : SYSMIS);
 }
 
 static bool
@@ -1644,7 +1625,9 @@ dump_statistics (const struct frq_proc *frq, const struct variable *wv)
           if (!pc->show)
             continue;
 
-          union value v = { .f = vf->tab.n_valid ? pc->value : SYSMIS };
+          union value v = {
+            .f = vf->tab.n_valid ? vf->percentiles[j] : SYSMIS
+          };
           pivot_table_put2 (table, var_idx, row++,
                             pivot_value_new_var_value (vf->var, &v));
         }