Categoricals: Replace single value by a case, so that interactions can be supported

diff --git a/src/language/stats/oneway.c b/src/language/stats/oneway.c
index 447bdad1b3f6ec6c7dc3a9daa7eb528f5c324925..9671328031d69f72ecd922bae0edc8dec0ffacdb 100644 (file)
--- a/src/language/stats/oneway.c
+++ b/src/language/stats/oneway.c
@@ -39,6 +39,7 @@
 #include "linreg/sweep.h"
 #include "tukey/tukey.h"
 #include "math/categoricals.h"
+#include "math/interaction.h"
 #include "math/covariance.h"
 #include "math/levene.h"
 #include "math/moments.h"
@@ -651,7 +652,7 @@ run_oneway (const struct oneway_spec *cmd,
 
   for (v = 0; v < cmd->n_vars; ++v)
     {
-      struct interaction *inter = interaction_create (cmd->indep_var);
+      const struct interaction *inter = interaction_create (cmd->indep_var);
       ws.vws[v].cat = categoricals_create (&inter, 1, cmd->wv,
                                            cmd->exclude, makeit, updateit,
                                            CONST_CAST (struct variable *,
@@ -1022,7 +1023,7 @@ show_descriptives (const struct oneway_spec *cmd, const struct oneway_workspace
 
          struct string vstr;
 
-         const union value *gval = categoricals_get_value_by_category (cats, count);
+         const struct ccase *gcc = categoricals_get_case_by_category (cats, count);
          const struct descriptive_data *dd = categoricals_get_user_data_by_category (cats, count);
 
          moments1_calculate (dd->mom, &n, &mean, &variance, NULL, NULL);
@@ -1032,7 +1033,7 @@ show_descriptives (const struct oneway_spec *cmd, const struct oneway_workspace
 
          ds_init_empty (&vstr);
 
-         var_append_value_name (cmd->indep_var, gval, &vstr);
+         var_append_value_name (cmd->indep_var, case_data (gcc, cmd->indep_var), &vstr);
 
          tab_text (t, 1, row + count,
                    TAB_LEFT | TAT_TITLE,
@@ -1232,13 +1233,13 @@ show_contrast_coeffs (const struct oneway_spec *cmd, const struct oneway_workspa
           ++count, coeffi = ll_next (coeffi))
        {
          const struct categoricals *cats = covariance_get_categoricals (cov);
-         const union value *val = categoricals_get_value_by_category (cats, count);
+         const struct ccase *gcc = categoricals_get_case_by_category (cats, count);
          struct coeff_node *coeffn = ll_data (coeffi, struct coeff_node, ll);
          struct string vstr;
 
          ds_init_empty (&vstr);
 
-         var_append_value_name (cmd->indep_var, val, &vstr);
+         var_append_value_name (cmd->indep_var, case_data (gcc, cmd->indep_var), &vstr);
 
          tab_text (t, count + 2, 1, TAB_CENTER | TAT_TITLE, ds_cstr (&vstr));
 
@@ -1519,10 +1520,11 @@ show_comparisons (const struct oneway_spec *cmd, const struct oneway_workspace *
          struct string vstr;
          int j;
          struct descriptive_data *dd_i = categoricals_get_user_data_by_category (cat, i);
-         const union value *gval = categoricals_get_value_by_category (cat, i);
+         const struct ccase *gcc = categoricals_get_case_by_category (cat, i);
+         
 
          ds_init_empty (&vstr);
-         var_append_value_name (cmd->indep_var, gval, &vstr);
+         var_append_value_name (cmd->indep_var, case_data (gcc, cmd->indep_var), &vstr);
 
          if ( i != 0)
            tab_hline (t, TAL_1, 1, n_cols - 1, r);
@@ -1540,8 +1542,8 @@ show_comparisons (const struct oneway_spec *cmd, const struct oneway_workspace *
                continue;
 
              ds_clear (&vstr);
-             gval = categoricals_get_value_by_category (cat, j);
-             var_append_value_name (cmd->indep_var, gval, &vstr);
+             const struct ccase *cc = categoricals_get_case_by_category (cat, j);
+             var_append_value_name (cmd->indep_var, case_data (cc, cmd->indep_var), &vstr);
              tab_text (t, 2, r + rx, TAB_LEFT | TAT_TITLE, ds_cstr (&vstr));
 
              moments1_calculate (dd_j->mom, &weight_j, &mean_j, &var_j, 0, 0);

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 2d65f8ae1902b5ad97f55e33602e173be4a6e19e..16bb46f6a1cbc72601c8e70b5e44c2f04508898f 100644 (file)
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -34,7 +34,7 @@
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
-  union value value;          /* The value being labeled. */
+  struct ccase *ccase;
   double cc;                  /* The total of the weights of cases with this value */
 
   void *user_data;            /* A pointer to data which the caller can store stuff */
@@ -49,6 +49,7 @@ struct var_params
   struct hmap map;
 
   const struct variable *var;
+  const struct interaction *iact;
 
   int base_subscript_short;
   int base_subscript_long;
@@ -72,7 +73,7 @@ compare_value_node (const void *vn1_, const void *vn2_, const void *aux)
   const struct value_node * const *vn2 = vn2_;
   const struct var_params *vp = aux;
 
-  return value_compare_3way (&(*vn1)->value, &(*vn2)->value, var_get_width (vp->var));
+  return interaction_case_cmp_3way (vp->iact, (*vn1)->ccase, (*vn2)->ccase);
 }
 
 
@@ -195,23 +196,21 @@ categoricals_dump (const struct categoricals *cat)
 #endif
 
 
-
 static struct value_node *
-lookup_value (const struct hmap *map, const struct variable *var, const union value *val)
+lookup_case (const struct hmap *map, const struct interaction *iact, const struct ccase *c)
 {
-  struct value_node *foo;
-  unsigned int width = var_get_width (var);
-  size_t hash = value_hash (val, width, 0);
+  struct value_node *nn;
+  size_t hash = interaction_case_hash (iact, c);
 
-  HMAP_FOR_EACH_WITH_HASH (foo, struct value_node, node, hash, map)
+  HMAP_FOR_EACH_WITH_HASH (nn, struct value_node, node, hash, map)
     {
-      if (value_equal (val, &foo->value, width))
+      if (interaction_case_equal (iact, c, nn->ccase))
        break;
 
       fprintf (stderr, "Warning: Hash table collision\n");
     }
 
-  return foo;
+  return nn;
 }
 
 
@@ -246,6 +245,7 @@ categoricals_create (const struct interaction **inter, size_t n_inter,
     {
       hmap_init (&cat->vp[i].map);
       cat->vp[i].var = inter[i]->vars[0];
+      cat->vp[i].iact = inter[i];
     }
 
   return cat;
@@ -265,24 +265,27 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 
   for (i = 0 ; i < cat->n_vp; ++i)
     {
+      const struct interaction *iact = cat->vp[i].iact;
       const struct variable *var = cat->vp[i].var;
-      unsigned int width = var_get_width (var);
+
       const union value *val = case_data (c, var);
       size_t hash;
       struct value_node *node ;
 
+#if XXX
       if ( var_is_value_missing (var, val, cat->exclude))
        continue;
+#endif
 
-      hash = value_hash (val, width, 0);
-      node = lookup_value (&cat->vp[i].map, var, val);
+      hash = interaction_case_hash (iact, c);
+      node = lookup_case (&cat->vp[i].map, iact, c);
 
       if ( NULL == node)
        {
+         int width  = var_get_width (var);
          node = pool_malloc (cat->pool, sizeof *node);
 
-         value_init (&node->value, width);
-         value_copy (&node->value, val, width);
+         node->ccase = case_ref (c);
          node->cc = 0.0;
 
          hmap_insert (&cat->vp[i].map, &node->node,  hash);
@@ -409,15 +412,25 @@ categoricals_get_variable_by_subscript (const struct categoricals *cat, int subs
   return cat->vp[index].var;
 }
 
-/* Return the value corresponding to SUBSCRIPT */
-static const union value *
-categoricals_get_value_by_subscript (const struct categoricals *cat, int subscript)
+/* Return the interaction corresponding to SUBSCRIPT */
+static const struct interaction *
+categoricals_get_interaction_by_subscript (const struct categoricals *cat, int subscript)
+{
+  int index = reverse_variable_lookup_short (cat, subscript);
+
+  return cat->vp[index].iact;
+}
+
+
+/* Return the case corresponding to SUBSCRIPT */
+static const struct ccase *
+categoricals_get_case_by_subscript (const struct categoricals *cat, int subscript)
 {
   int vindex = reverse_variable_lookup_short (cat, subscript);
   const struct var_params *vp = &cat->vp[vindex];
   const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_short];
 
-  return &vn->value;
+  return vn->ccase;
 }
 
 
@@ -447,12 +460,11 @@ double
 categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscript,
                                      const struct ccase *c)
 {
-  const struct variable *var = categoricals_get_variable_by_subscript (cat, subscript);
-  int width = var_get_width (var);
+  const struct interaction *iact = categoricals_get_interaction_by_subscript (cat, subscript);
 
-  const union value *val = case_data (c, var);
+  const struct ccase *c2 =  categoricals_get_case_by_subscript (cat, subscript);
 
-  return value_equal (val, categoricals_get_value_by_subscript (cat, subscript), width);
+  return interaction_case_equal (iact, c, c2);
 }
 
 
@@ -463,16 +475,15 @@ categoricals_get_n_variables (const struct categoricals *cat)
 }
 
 
-
-/* Return the value corresponding to SUBSCRIPT */
-const union value *
-categoricals_get_value_by_category (const struct categoricals *cat, int subscript)
+/* Return a case containing the set of values corresponding to SUBSCRIPT */
+const struct ccase *
+categoricals_get_case_by_category (const struct categoricals *cat, int subscript)
 {
   int vindex = reverse_variable_lookup_long (cat, subscript);
   const struct var_params *vp = &cat->vp[vindex];
   const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_long];
 
-  return &vn->value;
+  return vn->ccase;
 }
 
 

diff --git a/src/math/categoricals.h b/src/math/categoricals.h
index 6ab451d73949ebcc723df4f577b99749b6e97476..2cd55b5ec3afa1a6d8630d754bf682518c817c56 100644 (file)
--- a/src/math/categoricals.h
+++ b/src/math/categoricals.h
@@ -93,6 +93,7 @@ const union value * categoricals_get_value_by_category (const struct categorical
 
 void * categoricals_get_user_data_by_category (const struct categoricals *cat, int category);
 
+const struct ccase * categoricals_get_case_by_category (const struct categoricals *cat, int subscript);
 
 
 #endif

diff --git a/src/math/interaction.c b/src/math/interaction.c
index aa24250007da27b09723f5ddc4f115419fb10db5..afdc3277754884a69fe83b090c2c380321d55935 100644 (file)
--- a/src/math/interaction.c
+++ b/src/math/interaction.c
@@ -130,6 +130,25 @@ interaction_case_equal (const struct interaction *iact, const struct ccase *c1,
   return same;
 }
 
+
+int
+interaction_case_cmp_3way (const struct interaction *iact, const struct ccase *c1, const struct ccase *c2)
+{
+  int i;
+  int result = 0;
+
+  for (i = 0; i < iact->n_vars; ++i)
+    {
+      const struct variable *var = iact->vars[i];
+      result = value_compare_3way (case_data (c1, var), case_data (c2, var), var_get_width (var));
+      if (result != 0)
+       break;
+    }
+
+  return result;
+}
+
+
 bool
 interaction_case_is_missing (const struct interaction *iact, const struct ccase *c, enum mv_class exclude)
 {

diff --git a/src/math/interaction.h b/src/math/interaction.h
index 96c2940aef72d7e1f13f1305c9112f58d347fa44..cd38ae226ea81b6e288c80c6a4aa30aa03e8e066 100644 (file)
--- a/src/math/interaction.h
+++ b/src/math/interaction.h
@@ -43,5 +43,7 @@ struct ccase;
 unsigned int interaction_case_hash (const struct interaction *, const struct ccase *);
 bool interaction_case_equal (const struct interaction *, const struct ccase *, const struct ccase *);
 bool interaction_case_is_missing (const struct interaction *, const struct ccase *, enum mv_class);
+int interaction_case_cmp_3way (const struct interaction *, const struct ccase *, const struct ccase *);
+
 
 #endif