simplify weighting
authorBen Pfaff <blp@cs.stanford.edu>
Thu, 25 Aug 2022 20:18:58 +0000 (13:18 -0700)
committerBen Pfaff <blp@cs.stanford.edu>
Thu, 25 Aug 2022 20:18:58 +0000 (13:18 -0700)
src/language/stats/ctables.c
src/language/stats/ctables.inc
tests/language/stats/ctables.at

index b5152ef6365bedfdc1791010ce9ada75793c6382..7e899d1b668a640355ead9e7ccf83fd8c340490f 100644 (file)
@@ -63,18 +63,32 @@ enum ctables_vlabel
     CTVL_BOTH = SETTINGS_VALUE_SHOW_BOTH,
   };
 
+enum ctables_weighting
+  {
+    CTW_EFFECTIVE,
+    CTW_DICTIONARY,
+    CTW_UNWEIGHTED
+#define N_CTWS 3
+  };
+
 enum ctables_function_type
   {
-    /* A function that operates on data in a single cell.  The function does
-       not have an unweighted version. */
+    /* A function that operates on data in a single cell.  It operates on
+       effective weights.  It does not have an unweighted version. */
     CTFT_CELL,
 
-    /* A function that operates on data in a single cell.  The function has an
-       unweighted version. */
+    /* A function that operates on data in a single cell.  The function
+       operates on effective weights and has a U-prefixed unweighted
+       version. */
     CTFT_UCELL,
 
-    /* A function that operates on an area of cells.  The function has an
-       unweighted version. */
+    /* A function that operates on data in a single cell.  It operates on
+       dictionary weights, and has U-prefixed unweighted version and an
+       E-prefixed effective weight version. */
+    CTFT_UECELL,
+
+    /* A function that operates on an area of cells.  It operates on effective
+       weights and has a U-prefixed unweighted version. */
     CTFT_AREA,
   };
 
@@ -113,8 +127,9 @@ struct ctables_function_info
     enum ctables_format format;
     enum ctables_function_availability availability;
 
-    bool may_be_unweighted;
-    bool is_area;
+    bool u_prefix;              /* Accepts a 'U' prefix (for unweighted)? */
+    bool e_prefix;              /* Accepts an 'E' prefix (for effective)? */
+    bool is_area;               /* Needs an area prefix. */
   };
 static const struct ctables_function_info ctables_function_info[N_CTSF_FUNCTIONS] = {
 #define S(ENUM, NAME, TYPE, FORMAT, AVAILABILITY)                       \
@@ -123,15 +138,14 @@ static const struct ctables_function_info ctables_function_info[N_CTSF_FUNCTIONS
     .type = TYPE,                                                       \
     .format = FORMAT,                                                   \
     .availability = AVAILABILITY,                                       \
-    .may_be_unweighted = (TYPE) == CTFT_UCELL || (TYPE) == CTFT_AREA,   \
+    .u_prefix = (TYPE) == CTFT_UCELL || (TYPE) == CTFT_UECELL || (TYPE) == CTFT_AREA, \
+    .e_prefix = (TYPE) == CTFT_UECELL,                                  \
     .is_area = (TYPE) == CTFT_AREA                                      \
   },
 #include "ctables.inc"
 #undef S
 };
 
-static bool ctables_summary_function_is_count (enum ctables_summary_function);
-
 enum ctables_area_type
   {
     /* Within a section, where stacked variables divide one section from
@@ -169,22 +183,15 @@ struct ctables_area
     const struct ctables_cell *example;
 
     size_t sequence;
-    double d_valid;             /* Dictionary weight. */
-    double d_count;
-    double d_total;
-    double e_valid;             /* Effective weight */
-    double e_count;
-    double e_total;
-    double u_valid;             /* Unweighted. */
-    double u_count;
-    double u_total;
+    double count[N_CTWS];
+    double valid[N_CTWS];
+    double total[N_CTWS];
     struct ctables_sum *sums;
   };
 
 struct ctables_sum
   {
-    double e_sum;
-    double u_sum;
+    double sum[N_CTWS];
   };
 
 enum ctables_summary_variant
@@ -543,7 +550,7 @@ struct ctables_category
 
             /* CCT_FUNCTION. */
             enum ctables_summary_function sort_function;
-            bool weighted;
+            enum ctables_weighting weighting;
             enum ctables_area_type area;
             struct variable *sort_var;
             double percentile;
@@ -746,7 +753,7 @@ struct ctables_summary_spec
        cell, it must be 0).  For CTSF_PTILE only, 'percentile' is the
        percentile between 0 and 100 (for other functions it must be 0). */
     enum ctables_summary_function function;
-    bool weighted;
+    enum ctables_weighting weighting;
     enum ctables_area_type calc_area;
     double percentile;          /* CTSF_PTILE only. */
 
@@ -856,23 +863,18 @@ ctables_function_availability (enum ctables_summary_function f)
   return availability[f];
 }
 
-static bool
-ctables_summary_function_is_count (enum ctables_summary_function f)
-{
-  return f == CTSF_COUNT || f == CTSF_ECOUNT;
-}
-
 static bool
 parse_ctables_summary_function (struct lexer *lexer,
                                 enum ctables_summary_function *function,
-                                bool *weighted,
+                                enum ctables_weighting *weighting,
                                 enum ctables_area_type *area)
 {
   if (!lex_force_id (lexer))
     return false;
 
   struct substring name = lex_tokss (lexer);
-  *weighted = !(ss_match_byte (&name, 'U') || ss_match_byte (&name, 'u'));
+  bool u = ss_match_byte (&name, 'U') || ss_match_byte (&name, 'u');
+  bool e = !u && (ss_match_byte (&name, 'E') || ss_match_byte (&name, 'e'));
 
   bool has_area = false;
   *area = 0;
@@ -886,6 +888,7 @@ parse_ctables_summary_function (struct lexer *lexer,
           {
             /* Special case where .COUNT suffix is omitted. */
             *function = CTSF_areaPCT_COUNT;
+            *weighting = CTW_EFFECTIVE;
             lex_get (lexer);
             return true;
           }
@@ -898,11 +901,13 @@ parse_ctables_summary_function (struct lexer *lexer,
       if (ss_equals_case (cfi->basename, name))
         {
           *function = f;
-          if (!*weighted && !cfi->may_be_unweighted)
-            break;
-          if (has_area != cfi->is_area)
+          if ((u && !cfi->u_prefix) || (e && !cfi->e_prefix) || (has_area != cfi->is_area))
             break;
 
+          *weighting = (e ? CTW_EFFECTIVE
+                        : u ? CTW_UNWEIGHTED
+                        : cfi->e_prefix ? CTW_DICTIONARY
+                        : CTW_EFFECTIVE);
           lex_get (lexer);
           return true;
         }
@@ -986,15 +991,15 @@ ctables_summary_default_format (enum ctables_summary_function function,
 static const char *
 ctables_summary_label__ (const struct ctables_summary_spec *spec)
 {
-  bool w = spec->weighted;
+  bool w = spec->weighting != CTW_UNWEIGHTED;
+  bool d = spec->weighting == CTW_DICTIONARY;
   enum ctables_area_type a = spec->user_area;
   switch (spec->function)
     {
     case CTSF_COUNT:
-      return w ? N_("Count") : N_("Unweighted Count");
-
-    case CTSF_ECOUNT:
-      return N_("Adjusted Count");
+      return (d ? N_("Count")
+              : w ? N_("Adjusted Count")
+              : N_("Unweighted Count"));
 
     case CTSF_areaPCT_COUNT:
       switch (a)
@@ -1037,20 +1042,22 @@ ctables_summary_label__ (const struct ctables_summary_spec *spec)
 
     case CTSF_MAXIMUM: return N_("Maximum");
     case CTSF_MEAN: return w ? N_("Mean") : N_("Unweighted Mean");
-    case CTSF_MEDIAN: return N_("Median");
+    case CTSF_MEDIAN: return w ? N_("Median") : N_("Unweighted Median");
     case CTSF_MINIMUM: return N_("Minimum");
-    case CTSF_MISSING: return N_("Missing");
-    case CTSF_MODE: return N_("Mode");
+    case CTSF_MISSING: return w ? N_("Missing") : N_("Unweighted Missing");
+    case CTSF_MODE: return w ? N_("Mode") : N_("Unweighted Mode");
     case CTSF_PTILE: NOT_REACHED ();
     case CTSF_RANGE: return N_("Range");
-    case CTSF_SEMEAN: return N_("Std Error of Mean");
-    case CTSF_STDDEV: return N_("Std Deviation");
-    case CTSF_SUM: return N_("Sum");
-    case CTSF_TOTALN: return N_("Total N");
-    case CTSF_ETOTALN: return N_("Adjusted Total N");
-    case CTSF_VALIDN: return N_("Valid N");
-    case CTSF_EVALIDN: return N_("Adjusted Valid N");
-    case CTSF_VARIANCE: return N_("Variance");
+    case CTSF_SEMEAN: return w ? N_("Std Error of Mean") : N_("Unweighted Std Error of Mean");
+    case CTSF_STDDEV: return w ? N_("Std Deviation") : N_("Unweighted Std Deviation");
+    case CTSF_SUM: return w ? N_("Sum") : N_("Unweighted Sum");
+    case CTSF_TOTALN: return (d ? N_("Total N")
+                              : w ? N_("Adjusted Total N")
+                              : N_("Unweighted Total N"));
+    case CTSF_VALIDN: return (d ? N_("Valid N")
+                              : w ? N_("Adjusted Valid N")
+                              : N_("Unweighted Valid N"));
+    case CTSF_VARIANCE: return w ? N_("Variance") : N_("Unweighted Variance");
     case CTSF_areaPCT_SUM:
       switch (a)
         {
@@ -1090,7 +1097,7 @@ ctables_summary_label (const struct ctables_summary_spec *spec, double cilevel)
       if (spec->function == CTSF_PTILE)
         {
           double p = spec->percentile;
-          char *s = (spec->weighted
+          char *s = (spec->weighting != CTW_UNWEIGHTED
                      ? xasprintf (_("Percentile %.2f"), p)
                      : xasprintf (_("Unweighted Percentile %.2f"), p));
           return pivot_value_new_user_text_nocopy (s);
@@ -1120,13 +1127,16 @@ ctables_summary_label (const struct ctables_summary_spec *spec, double cilevel)
 
 static const char *
 ctables_summary_function_name (enum ctables_summary_function function,
-                               bool weighted,
+                               enum ctables_weighting weighting,
                                enum ctables_area_type area,
                                char *buffer, size_t bufsize)
 {
   const struct ctables_function_info *cfi = &ctables_function_info[function];
   snprintf (buffer, bufsize, "%s%s%s",
-            weighted ? "" : "U",
+            (weighting == CTW_UNWEIGHTED ? "U"
+             : weighting == CTW_DICTIONARY ? ""
+             : cfi->e_prefix ? "E"
+             : ""),
             cfi->is_area ? ctables_area_type_name[area] : "",
             cfi->basename.string);
   return buffer;
@@ -1134,7 +1144,8 @@ ctables_summary_function_name (enum ctables_summary_function function,
 
 static bool
 add_summary_spec (struct ctables_axis *axis,
-                  enum ctables_summary_function function, bool weighted,
+                  enum ctables_summary_function function,
+                  enum ctables_weighting weighting,
                   enum ctables_area_type area, double percentile,
                   const char *label, const struct fmt_spec *format,
                   bool is_ctables_format, const struct msg_location *loc,
@@ -1143,7 +1154,7 @@ add_summary_spec (struct ctables_axis *axis,
   if (axis->op == CTAO_VAR)
     {
       char function_name[128];
-      ctables_summary_function_name (function, weighted, area,
+      ctables_summary_function_name (function, weighting, area,
                                      function_name, sizeof function_name);
       const char *var_name = var_get_name (axis->var);
       switch (ctables_function_availability (function))
@@ -1181,7 +1192,7 @@ add_summary_spec (struct ctables_axis *axis,
       struct ctables_summary_spec *dst = &set->specs[set->n++];
       *dst = (struct ctables_summary_spec) {
         .function = function,
-        .weighted = weighted,
+        .weighting = weighting,
         .calc_area = area,
         .user_area = area,
         .percentile = percentile,
@@ -1195,7 +1206,7 @@ add_summary_spec (struct ctables_axis *axis,
   else
     {
       for (size_t i = 0; i < 2; i++)
-        if (!add_summary_spec (axis->subs[i], function, weighted, area,
+        if (!add_summary_spec (axis->subs[i], function, weighting, area,
                                percentile, label, format, is_ctables_format,
                                loc, sv))
           return false;
@@ -1313,9 +1324,9 @@ ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
 
       /* Parse function. */
       enum ctables_summary_function function;
-      bool weighted;
+      enum ctables_weighting weighting;
       enum ctables_area_type area;
-      if (!parse_ctables_summary_function (ctx->lexer, &function, &weighted,
+      if (!parse_ctables_summary_function (ctx->lexer, &function, &weighting,
                                            &area))
         goto error;
 
@@ -1357,7 +1368,7 @@ ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
 
       struct msg_location *loc = lex_ofs_location (ctx->lexer, start_ofs,
                                                    lex_ofs (ctx->lexer) - 1);
-      add_summary_spec (sub, function, weighted, area, percentile, label,
+      add_summary_spec (sub, function, weighting, area, percentile, label,
                         formatp, is_ctables_format, loc, sv);
       free (label);
       msg_location_destroy (loc);
@@ -2107,7 +2118,7 @@ ctables_table_parse_categories (struct lexer *lexer, struct dictionary *dict,
             {
               cat.type = CCT_FUNCTION;
               if (!parse_ctables_summary_function (lexer, &cat.sort_function,
-                                                   &cat.weighted, &cat.area))
+                                                   &cat.weighting, &cat.area))
                 goto error;
 
               if (lex_match (lexer, T_LPAREN))
@@ -2535,15 +2546,12 @@ ctables_summary_init (union ctables_summary *s,
   switch (ss->function)
     {
     case CTSF_COUNT:
-    case CTSF_ECOUNT:
     case CTSF_areaPCT_COUNT:
     case CTSF_areaPCT_VALIDN:
     case CTSF_areaPCT_TOTALN:
     case CTSF_MISSING:
     case CTSF_TOTALN:
-    case CTSF_ETOTALN:
     case CTSF_VALIDN:
-    case CTSF_EVALIDN:
       s->count = 0;
       break;
 
@@ -2596,15 +2604,12 @@ ctables_summary_uninit (union ctables_summary *s,
   switch (ss->function)
     {
     case CTSF_COUNT:
-    case CTSF_ECOUNT:
     case CTSF_areaPCT_COUNT:
     case CTSF_areaPCT_VALIDN:
     case CTSF_areaPCT_TOTALN:
     case CTSF_MISSING:
     case CTSF_TOTALN:
-    case CTSF_ETOTALN:
     case CTSF_VALIDN:
-    case CTSF_EVALIDN:
       break;
 
     case CTSF_areaID:
@@ -2638,7 +2643,7 @@ ctables_summary_add (union ctables_summary *s,
                      const struct variable *var, const union value *value,
                      bool is_scale, bool is_scale_missing,
                      bool is_missing, bool is_included,
-                     double d_weight, double e_weight)
+                     double weight)
 {
   /* To determine whether a case is included in a given table for a particular
      kind of summary, consider the following charts for each variable in the
@@ -2661,35 +2666,35 @@ ctables_summary_add (union ctables_summary *s,
   switch (ss->function)
     {
     case CTSF_TOTALN:
-      s->count += ss->weighted ? d_weight : 1.0;
+      s->count += weight;
       break;
 
     case CTSF_areaPCT_TOTALN:
-      s->count += ss->weighted ? e_weight : 1.0;
+      s->count += weight;
       break;
 
     case CTSF_COUNT:
       if (is_scale || is_included)
-        s->count += ss->weighted ? d_weight : 1.0;
+        s->count += weight;
       break;
 
     case CTSF_areaPCT_COUNT:
       if (is_scale || is_included)
-        s->count += ss->weighted ? e_weight : 1.0;
+        s->count += weight;
       break;
 
     case CTSF_VALIDN:
       if (is_scale
           ? !is_scale_missing
           : !is_missing)
-        s->count += ss->weighted ? d_weight : 1.0;
+        s->count += weight;
       break;
 
     case CTSF_areaPCT_VALIDN:
       if (is_scale
           ? !is_scale_missing
           : !is_missing)
-        s->count += ss->weighted ? e_weight : 1.0;
+        s->count += weight;
       break;
 
     case CTSF_areaID:
@@ -2699,23 +2704,7 @@ ctables_summary_add (union ctables_summary *s,
       if (is_scale
           ? is_scale_missing
           : is_missing)
-        s->count += ss->weighted ? e_weight : 1.0;
-      break;
-
-    case CTSF_ECOUNT:
-      if (is_scale || is_included)
-        s->count += e_weight;
-      break;
-
-    case CTSF_EVALIDN:
-      if (is_scale
-          ? !is_scale_missing
-          : !is_missing)
-        s->count += e_weight;
-      break;
-
-    case CTSF_ETOTALN:
-      s->count += e_weight;
+        s->count += weight;
       break;
 
     case CTSF_MAXIMUM:
@@ -2737,12 +2726,12 @@ ctables_summary_add (union ctables_summary *s,
     case CTSF_SUM:
     case CTSF_VARIANCE:
       if (!is_scale_missing)
-        moments1_add (s->moments, value->f, ss->weighted ? e_weight : 1.0);
+        moments1_add (s->moments, value->f, weight);
       break;
 
     case CTSF_areaPCT_SUM:
       if (!is_missing && !is_scale_missing)
-        moments1_add (s->moments, value->f, ss->weighted ? e_weight : 1.0);
+        moments1_add (s->moments, value->f, weight);
       break;
 
     case CTSF_MEDIAN:
@@ -2750,12 +2739,11 @@ ctables_summary_add (union ctables_summary *s,
     case CTSF_PTILE:
       if (!is_scale_missing)
         {
-          double w = ss->weighted ? e_weight : 1.0;
-          s->ovalid += w;
+          s->ovalid += weight;
 
           struct ccase *c = case_create (casewriter_get_proto (s->writer));
           *case_num_rw_idx (c, 0) = value->f;
-          *case_num_rw_idx (c, 1) = w;
+          *case_num_rw_idx (c, 1) = weight;
           casewriter_write (s->writer, c);
         }
       break;
@@ -2770,7 +2758,6 @@ ctables_summary_value (const struct ctables_cell *cell,
   switch (ss->function)
     {
     case CTSF_COUNT:
-    case CTSF_ECOUNT:
       return s->count;
 
     case CTSF_areaID:
@@ -2779,29 +2766,27 @@ ctables_summary_value (const struct ctables_cell *cell,
     case CTSF_areaPCT_COUNT:
       {
         const struct ctables_area *a = cell->areas[ss->calc_area];
-        double a_count = ss->weighted ? a->e_count : a->u_count;
+        double a_count = a->count[ss->weighting];
         return a_count ? s->count / a_count * 100 : SYSMIS;
       }
 
     case CTSF_areaPCT_VALIDN:
       {
         const struct ctables_area *a = cell->areas[ss->calc_area];
-        double a_valid = ss->weighted ? a->e_valid : a->u_valid;
+        double a_valid = a->valid[ss->weighting];
         return a_valid ? s->count / a_valid * 100 : SYSMIS;
       }
 
     case CTSF_areaPCT_TOTALN:
       {
         const struct ctables_area *a = cell->areas[ss->calc_area];
-        double a_total = ss->weighted ? a->e_total : a->u_total;
+        double a_total = a->total[ss->weighting];
         return a_total ? s->count / a_total * 100 : SYSMIS;
       }
 
     case CTSF_MISSING:
     case CTSF_TOTALN:
-    case CTSF_ETOTALN:
     case CTSF_VALIDN:
-    case CTSF_EVALIDN:
       return s->count;
 
     case CTSF_MAXIMUM:
@@ -2857,7 +2842,7 @@ ctables_summary_value (const struct ctables_cell *cell,
 
         const struct ctables_area *a = cell->areas[ss->calc_area];
         const struct ctables_sum *sum = &a->sums[ss->sum_var_idx];
-        double denom = ss->weighted ? sum->e_sum : sum->u_sum;
+        double denom = sum->sum[ss->weighting];
         return denom != 0 ? weight * mean / denom * 100 : SYSMIS;
       }
 
@@ -3319,10 +3304,17 @@ is_listwise_missing (const struct ctables_summary_spec_set *specs,
   return false;
 }
 
+static void
+add_weight (double dst[N_CTWS], const double src[N_CTWS])
+{
+  for (enum ctables_weighting wt = 0; wt < N_CTWS; wt++)
+    dst[wt] += src[wt];
+}
+
 static void
 ctables_cell_add__ (struct ctables_section *s, const struct ccase *c,
                     const struct ctables_category *cats[PIVOT_N_AXES][10],
-                    bool is_included, double d_weight, double e_weight)
+                    bool is_included, double weight[N_CTWS])
 {
   struct ctables_cell *cell = ctables_cell_insert__ (s, c, cats);
   const struct ctables_nest *ss = s->nests[s->table->summary_axis];
@@ -3334,27 +3326,20 @@ ctables_cell_add__ (struct ctables_section *s, const struct ccase *c,
 
   for (size_t i = 0; i < specs->n; i++)
      ctables_summary_add (&cell->summaries[i], &specs->specs[i],
-                         specs->var, value, specs->is_scale,
-                         scale_missing, is_missing, is_included,
-                         d_weight, e_weight);
+                          specs->var, value, specs->is_scale,
+                          scale_missing, is_missing, is_included,
+                          weight[specs->specs[i].weighting]);
   for (enum ctables_area_type at = 0; at < N_CTATS; at++)
     if (!(cell->omit_areas && (1u << at)))
       {
         struct ctables_area *a = cell->areas[at];
-        a->d_total += d_weight;
-        a->e_total += e_weight;
-        a->u_total += 1.0;
+
+        add_weight (a->total, weight);
         if (is_included)
-          {
-            a->d_count += d_weight;
-            a->e_count += e_weight;
-            a->u_count += 1.0;
-          }
+          add_weight (a->count, weight);
         if (!is_missing)
           {
-            a->d_valid += d_weight;
-            a->e_valid += e_weight;
-            a->u_count += 1.0;
+            add_weight (a->valid, weight);
 
             for (size_t i = 0; i < s->table->n_sum_vars; i++)
               {
@@ -3364,8 +3349,8 @@ ctables_cell_add__ (struct ctables_section *s, const struct ccase *c,
                 if (!var_is_num_missing (var, addend))
                   {
                     struct ctables_sum *sum = &a->sums[i];
-                    sum->e_sum += addend * e_weight;
-                    sum->u_sum += addend;
+                    for (enum ctables_weighting wt = 0; wt < N_CTWS; wt++)
+                      sum->sum[wt] += addend * weight[wt];
                   }
               }
           }
@@ -3375,7 +3360,7 @@ ctables_cell_add__ (struct ctables_section *s, const struct ccase *c,
 static void
 recurse_totals (struct ctables_section *s, const struct ccase *c,
                 const struct ctables_category *cats[PIVOT_N_AXES][10],
-                bool is_included, double d_weight, double e_weight,
+                bool is_included, double weight[N_CTWS],
                 enum pivot_axis_type start_axis, size_t start_nest)
 {
   for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
@@ -3394,8 +3379,8 @@ recurse_totals (struct ctables_section *s, const struct ccase *c,
             {
               const struct ctables_category *save = cats[a][i];
               cats[a][i] = total;
-              ctables_cell_add__ (s, c, cats, is_included, d_weight, e_weight);
-              recurse_totals (s, c, cats, is_included, d_weight, e_weight, a, i + 1);
+              ctables_cell_add__ (s, c, cats, is_included, weight);
+              recurse_totals (s, c, cats, is_included, weight, a, i + 1);
               cats[a][i] = save;
             }
         }
@@ -3406,7 +3391,7 @@ recurse_totals (struct ctables_section *s, const struct ccase *c,
 static void
 recurse_subtotals (struct ctables_section *s, const struct ccase *c,
                    const struct ctables_category *cats[PIVOT_N_AXES][10],
-                   bool is_included, double d_weight, double e_weight,
+                   bool is_included, double weight[N_CTWS],
                    enum pivot_axis_type start_axis, size_t start_nest)
 {
   for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
@@ -3421,8 +3406,8 @@ recurse_subtotals (struct ctables_section *s, const struct ccase *c,
           if (save->subtotal)
             {
               cats[a][i] = save->subtotal;
-              ctables_cell_add__ (s, c, cats, is_included, d_weight, e_weight);
-              recurse_subtotals (s, c, cats, is_included, d_weight, e_weight, a, i + 1);
+              ctables_cell_add__ (s, c, cats, is_included, weight);
+              recurse_subtotals (s, c, cats, is_included, weight, a, i + 1);
               cats[a][i] = save;
             }
         }
@@ -3450,9 +3435,8 @@ ctables_add_occurrence (const struct variable *var,
 }
 
 static void
-ctables_cell_insert (struct ctables_section *s,
-                     const struct ccase *c,
-                     double d_weight, double e_weight)
+ctables_cell_insert (struct ctables_section *s, const struct ccase *c,
+                     double weight[N_CTWS])
 {
   const struct ctables_category *cats[PIVOT_N_AXES][10]; /* XXX */
 
@@ -3500,9 +3484,9 @@ ctables_cell_insert (struct ctables_section *s,
             }
       }
 
-  ctables_cell_add__ (s, c, cats, is_included, d_weight, e_weight);
-  recurse_totals (s, c, cats, is_included, d_weight, e_weight, 0, 0);
-  recurse_subtotals (s, c, cats, is_included, d_weight, e_weight, 0, 0);
+  ctables_cell_add__ (s, c, cats, is_included, weight);
+  recurse_totals (s, c, cats, is_included, weight, 0, 0);
+  recurse_subtotals (s, c, cats, is_included, weight, 0, 0);
 }
 
 struct merge_item
@@ -3518,8 +3502,8 @@ merge_item_compare_3way (const struct merge_item *a, const struct merge_item *b)
   const struct ctables_summary_spec *bs = &b->set->specs[b->ofs];
   if (as->function != bs->function)
     return as->function > bs->function ? 1 : -1;
-  else if (as->weighted != bs->weighted)
-    return as->weighted > bs->weighted ? 1 : -1;
+  else if (as->weighting != bs->weighting)
+    return as->weighting > bs->weighting ? 1 : -1;
   else if (as->calc_area != bs->calc_area)
     return as->calc_area > bs->calc_area ? 1 : -1;
   else if (as->percentile != bs->percentile)
@@ -4016,7 +4000,7 @@ ctables_cell_calculate_postcompute (const struct ctables_section *s,
         {
           const struct ctables_summary_spec *ss2 = &pc->specs->specs[i];
           if (ss->function == ss2->function
-              && ss->weighted == ss2->weighted
+              && ss->weighting == ss2->weighting
               && ss->calc_area == ss2->calc_area
               && ss->percentile == ss2->percentile)
             {
@@ -4464,7 +4448,7 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t)
               struct pivot_value *value;
               if (ct->hide_threshold != 0
                   && d < ct->hide_threshold
-                  && ctables_summary_function_is_count (ss->function))
+                  && ss->function == CTSF_COUNT)
                 {
                   value = pivot_value_new_user_text_nocopy (
                     xasprintf ("<%d", ct->hide_threshold));
@@ -4786,7 +4770,7 @@ ctables_prepare_table (struct ctables_table *t)
             }
           *ss->specs = (struct ctables_summary_spec) {
             .function = function,
-            .weighted = true,
+            .weighting = ss->is_scale ? CTW_EFFECTIVE : CTW_DICTIONARY,
             .format = ctables_summary_default_format (function, ss->var),
           };
 
@@ -5269,13 +5253,18 @@ ctables_execute (struct dataset *ds, struct casereader *input,
                                                        case_num (c, ct->e_weight),
                                                        &warn_on_invalid)
                              : d_weight);
+          double weight[] = {
+            [CTW_DICTIONARY] = d_weight,
+            [CTW_EFFECTIVE] = e_weight,
+            [CTW_UNWEIGHTED] = 1.0,
+          };
 
           for (size_t i = 0; i < ct->n_tables; i++)
             {
               struct ctables_table *t = ct->tables[i];
 
               for (size_t j = 0; j < t->n_sections; j++)
-                ctables_cell_insert (&t->sections[j], c, d_weight, e_weight);
+                ctables_cell_insert (&t->sections[j], c, weight);
 
               for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
                 if (t->label_axis[a] != a)
@@ -5773,9 +5762,9 @@ ctables_parse_pproperties_format (struct lexer *lexer,
     {
       /* Parse function. */
       enum ctables_summary_function function;
-      bool weighted;
+      enum ctables_weighting weighting;
       enum ctables_area_type area;
-      if (!parse_ctables_summary_function (lexer, &function, &weighted, &area))
+      if (!parse_ctables_summary_function (lexer, &function, &weighting, &area))
         goto error;
 
       /* Parse percentile. */
@@ -5799,7 +5788,7 @@ ctables_parse_pproperties_format (struct lexer *lexer,
                                  sizeof *sss->specs);
       sss->specs[sss->n++] = (struct ctables_summary_spec) {
         .function = function,
-        .weighted = weighted,
+        .weighting = weighting,
         .calc_area = area,
         .user_area = area,
         .percentile = percentile,
index 1a418fe03d43b98d0d5ec6641b708385a05296e5..27160997e0960f17c2f3efbd8eeddba347c030ab 100644 (file)
@@ -1,30 +1,27 @@
 /* -*- c -*- */
 
 /* Summary functions for all variables. */
-S(CTSF_COUNT,          "COUNT",      CTFT_UCELL, CTF_COUNT,   CTFA_ALL)
-S(CTSF_ECOUNT,         "ECOUNT",     CTFT_CELL,  CTF_COUNT,   CTFA_ALL)
-S(CTSF_areaPCT_COUNT,  "PCT.COUNT",  CTFT_AREA,  CTF_PERCENT, CTFA_ALL)
-S(CTSF_areaPCT_VALIDN, "PCT.VALIDN", CTFT_AREA,  CTF_PERCENT, CTFA_ALL)
-S(CTSF_areaPCT_TOTALN, "PCT.TOTALN", CTFT_AREA,  CTF_PERCENT, CTFA_ALL)
+S(CTSF_COUNT,          "COUNT",      CTFT_UECELL, CTF_COUNT,   CTFA_ALL)
+S(CTSF_areaPCT_COUNT,  "PCT.COUNT",  CTFT_AREA,   CTF_PERCENT, CTFA_ALL)
+S(CTSF_areaPCT_VALIDN, "PCT.VALIDN", CTFT_AREA,   CTF_PERCENT, CTFA_ALL)
+S(CTSF_areaPCT_TOTALN, "PCT.TOTALN", CTFT_AREA,   CTF_PERCENT, CTFA_ALL)
 
 /* Scale variables, totals, and subtotals. */
-S(CTSF_MAXIMUM,        "MAXIMUM",    CTFT_CELL,  CTF_GENERAL, CTFA_SCALE)
-S(CTSF_MEAN,           "MEAN",       CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_MEDIAN,         "MEDIAN",     CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_MINIMUM,        "MINIMUM",    CTFT_CELL,  CTF_GENERAL, CTFA_SCALE)
-S(CTSF_MISSING,        "MISSING",    CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_MODE,           "MODE",       CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_PTILE,          "PTILE",      CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_RANGE,          "RANGE",      CTFT_CELL,  CTF_GENERAL, CTFA_SCALE)
-S(CTSF_SEMEAN,         "SEMEAN",     CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_STDDEV,         "STDDEV",     CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_SUM,            "SUM",        CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_TOTALN,         "TOTALN",     CTFT_UCELL, CTF_COUNT,   CTFA_SCALE)
-S(CTSF_ETOTALN,        "ETOTALN",    CTFT_CELL,  CTF_COUNT,   CTFA_SCALE)
-S(CTSF_VALIDN,         "VALIDN",     CTFT_UCELL, CTF_COUNT,   CTFA_SCALE)
-S(CTSF_EVALIDN,        "EVALIDN",    CTFT_CELL,  CTF_COUNT,   CTFA_SCALE)
-S(CTSF_VARIANCE,       "VARIANCE",   CTFT_UCELL, CTF_GENERAL, CTFA_SCALE)
-S(CTSF_areaPCT_SUM,    "PCT.SUM",    CTFT_AREA,  CTF_PERCENT, CTFA_SCALE)
+S(CTSF_MAXIMUM,        "MAXIMUM",    CTFT_CELL,   CTF_GENERAL, CTFA_SCALE)
+S(CTSF_MEAN,           "MEAN",       CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_MEDIAN,         "MEDIAN",     CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_MINIMUM,        "MINIMUM",    CTFT_CELL,   CTF_GENERAL, CTFA_SCALE)
+S(CTSF_MISSING,        "MISSING",    CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_MODE,           "MODE",       CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_PTILE,          "PTILE",      CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_RANGE,          "RANGE",      CTFT_CELL,   CTF_GENERAL, CTFA_SCALE)
+S(CTSF_SEMEAN,         "SEMEAN",     CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_STDDEV,         "STDDEV",     CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_SUM,            "SUM",        CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_TOTALN,         "TOTALN",     CTFT_UECELL, CTF_COUNT,   CTFA_SCALE)
+S(CTSF_VALIDN,         "VALIDN",     CTFT_UECELL, CTF_COUNT,   CTFA_SCALE)
+S(CTSF_VARIANCE,       "VARIANCE",   CTFT_UCELL,  CTF_GENERAL, CTFA_SCALE)
+S(CTSF_areaPCT_SUM,    "PCT.SUM",    CTFT_AREA,   CTF_PERCENT, CTFA_SCALE)
 
 /* Debugging and troubleshooting. */
-S(CTSF_areaID,         "ID",         CTFT_AREA,  CTF_COUNT,   CTFA_ALL)
+S(CTSF_areaID,         "ID",         CTFT_AREA,   CTF_COUNT,   CTFA_ALL)
index 40be19464b1bbb85951a333b4b637f6e1a2bf02e..67d6db4b263b97df81f8e819996b9b1a43deee8e 100644 (file)
@@ -5051,29 +5051,29 @@ AT_CHECK([pspp ctables.sps -O box=unicode -O width=120], [0], [dnl
 │  Layer Row Sum %  │  13.3%│ 10.3%│ 11.3%│  8.9%│  12.6%│ 11.5%│  9.8%│  9.9%│  12.4%│
 ╰───────────────────┴───────┴──────┴──────┴──────┴───────┴──────┴──────┴──────┴───────╯
 
-                                        Custom Tables
-╭────────────────────────────┬──────────────────────────────────────────────────────────────╮
-│                            │                               a                              │
-│                            ├────────────────────┬────────────────────┬────────────────────┤
-│                            │          1         │          2         │          9         │
-│                            ├────────────────────┼────────────────────┼────────────────────┤
-│                            │          b         │          b         │          b         │
-│                            ├──────┬──────┬──────┼──────┬──────┬──────┼──────┬──────┬──────┤
-│                            │   3  │   4  │   9  │   3  │   4  │   9  │   3  │   4  │   9  │
-├────────────────────────────┼──────┼──────┼──────┼──────┼──────┼──────┼──────┼──────┼──────┤
-│c Valid N                   │     7│     6│     8│     7│     7│     8│     7│     7│     8│
-│  Missing                   │     3│     4│     2│     3│     3│     2│     3│     3│     2│
-│  Unweighted Mean           │ 25.86│ 24.50│ 24.63│ 25.86│ 25.71│ 24.25│ 25.43│ 25.29│ 23.88│
-│  Std Error of Mean         │  2.44│  2.14│  2.58│  2.44│  2.18│  2.43│  2.36│  2.18│  2.47│
-│  Median                    │ 25.00│ 24.50│ 25.00│ 25.00│ 27.00│ 25.00│ 25.00│ 24.00│ 23.50│
-│  Mode                      │    16│    18│    15│    16│    18│    15│    16│    18│    15│
-│  Std Deviation             │  6.47│  5.24│  7.31│  6.47│  5.77│  6.88│  6.24│  5.77│  6.98│
-│  Variance                  │ 41.81│ 27.50│ 53.41│ 41.81│ 33.24│ 47.36│ 38.95│ 33.24│ 48.70│
-│  Sum                       │181.00│147.00│197.00│181.00│180.00│194.00│178.00│177.00│191.00│
-│  Unweighted Count          │    10│    10│    10│    10│    10│    10│    10│    10│    10│
-│  Total N                   │    10│    10│    10│    10│    10│    10│    10│    10│    10│
-│  Unweighted Layer Row Sum %│ 11.1%│  9.0%│ 12.1%│ 11.1%│ 11.1%│ 11.9%│ 10.9%│ 10.9%│ 11.7%│
-╰────────────────────────────┴──────┴──────┴──────┴──────┴──────┴──────┴──────┴──────┴──────╯
+                                         Custom Tables
\95­â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¬â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\95®
+│                              │                               a                              │
+│                              ├────────────────────┬────────────────────┬────────────────────┤
+│                              │          1         │          2         │          9         │
+│                              ├────────────────────┼────────────────────┼────────────────────┤
+│                              │          b         │          b         │          b         │
+│                              ├──────┬──────┬──────┼──────┬──────┬──────┼──────┬──────┬──────┤
+│                              │   3  │   4  │   9  │   3  │   4  │   9  │   3  │   4  │   9  │
\94\9câ\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¼â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94¤
+│c Unweighted Valid N          │     7│     6│     8│     7│     7│     8│     7│     7│     8│
+│  Unweighted Missing          │     3│     4│     2│     3│     3│     2│     3│     3│     2│
+│  Unweighted Mean             │ 25.86│ 24.50│ 24.63│ 25.86│ 25.71│ 24.25│ 25.43│ 25.29│ 23.88│
+│  Unweighted Std Error of Mean│  2.44│  2.14│  2.58│  2.44│  2.18│  2.43│  2.36│  2.18│  2.47│
+│  Unweighted Median           │ 25.00│ 24.50│ 25.00│ 25.00│ 27.00│ 25.00│ 25.00│ 24.00│ 23.50│
+│  Unweighted Mode             │    16│    18│    15│    16│    18│    15│    16│    18│    15│
+│  Unweighted Std Deviation    │  6.47│  5.24│  7.31│  6.47│  5.77│  6.88│  6.24│  5.77│  6.98│
+│  Unweighted Variance         │ 41.81│ 27.50│ 53.41│ 41.81│ 33.24│ 47.36│ 38.95│ 33.24│ 48.70│
+│  Unweighted Sum              │181.00│147.00│197.00│181.00│180.00│194.00│178.00│177.00│191.00│
+│  Unweighted Count            │    10│    10│    10│    10│    10│    10│    10│    10│    10│
+│  Unweighted Total N          │    10│    10│    10│    10│    10│    10│    10│    10│    10│
+│  Unweighted Layer Row Sum %  │ 11.1%│  9.0%│ 12.1%│ 11.1%│ 11.1%│ 11.9%│ 10.9%│ 10.9%│ 11.7%│
\95°â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\94´â\94\80â\94\80â\94\80â\94\80â\94\80â\94\80â\95¯
 ])
 AT_CLEANUP