Drop first category of each variable from covariance matrix.

[pspp-builds.git] / src / math / categoricals.c

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 4b5176322f798212073ffc6fab95e312789acf92..5c96c40cc359a1ec87f00594890eb00cfe75c577 100644 (file)
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -61,19 +61,29 @@ struct var_params
 
 struct categoricals
 {
+  /* The weight variable */
   const struct variable *wv;
 
-  size_t n_vars;
 
   /* An array of var_params */
   struct var_params *vp;
 
+  /* The size of VP. (ie, the number of variables involved.) */
+  size_t n_vp;
+
+  /* The number of categorical variables which contain entries.
+     In the absence of missing values, this will be equal to N_VP */
+  size_t n_vars;
+
   /* A map to enable the lookup of variables indexed by subscript */
   int *reverse_variable_map;
 
   size_t n_cats_total;
 
   struct pool *pool;
+
+  /* Missing values to be excluded */
+  enum mv_class exclude;
 };
 
 
@@ -81,7 +91,7 @@ void
 categoricals_destroy ( struct categoricals *cat)
 {
   int i;
-  for (i = 0 ; i < cat->n_vars; ++i)
+  for (i = 0 ; i < cat->n_vp; ++i)
     hmap_destroy (&cat->vp[i].map);
 
   pool_destroy (cat->pool);
@@ -94,17 +104,15 @@ categoricals_dump (const struct categoricals *cat)
 {
   int v;
 
-  for (v = 0 ; v < cat->n_vars; ++v)
+  for (v = 0 ; v < cat->n_vp; ++v)
     {
       const struct var_params *vp = &cat->vp[v];
       const struct hmap *m = &vp->map;
-      //      size_t width = var_get_width (vp->var);
       struct hmap_node *node ;
       int x;
      
-      printf ("\n%s (%d)  CC=%g:\n", var_get_name (vp->var), vp->base_subscript, vp->cc);
-
-      assert (vp->reverse_value_map);
+      printf ("\n%s (%d)  CC=%g n_cats=%d:\n", 
+             var_get_name (vp->var), vp->base_subscript, vp->cc, vp->n_cats);
 
       printf ("Reverse map\n");
       for (x = 0 ; x < vp->n_cats; ++x)
@@ -130,6 +138,19 @@ categoricals_dump (const struct categoricals *cat)
          ds_destroy (&s);
        }
     }
+
+  assert (cat->n_vars <= cat->n_vp);
+
+  printf ("\n");
+  printf ("Number of categorical variables: %d\n", cat->n_vp);
+  printf ("Number of non-empty categorical variables: %d\n", cat->n_vars);
+  printf ("Total number of categories: %d\n", cat->n_cats_total);
+
+  printf ("\nReverse variable map:\n");
+
+  for (v = 0 ; v < cat->n_cats_total - cat->n_vars; ++v)
+    printf ("%d ", cat->reverse_variable_map[v]);
+  printf ("\n");
 }
 
 
@@ -155,20 +176,23 @@ lookup_value (const struct hmap *map, const struct variable *var, const union va
 
 
 struct categoricals *
-categoricals_create (const struct variable **v, size_t n_vars, const struct variable *wv)
+categoricals_create (const struct variable **v, size_t n_vars,
+                    const struct variable *wv, enum mv_class exclude)
 {
   size_t i;
   struct categoricals *cat = xmalloc (sizeof *cat);
   
-  cat->n_vars = n_vars;
+  cat->n_vp = n_vars;
   cat->wv = wv;
   cat->n_cats_total = 0;
+  cat->n_vars = 0;
   cat->reverse_variable_map = NULL;
   cat->pool = pool_create ();
+  cat->exclude = exclude;
 
-  cat->vp = pool_calloc (cat->pool, n_vars, sizeof *cat->vp);
+  cat->vp = pool_calloc (cat->pool, cat->n_vp, sizeof *cat->vp);
 
-  for (i = 0 ; i < cat->n_vars; ++i)
+  for (i = 0 ; i < cat->n_vp; ++i)
     {
       hmap_init (&cat->vp[i].map);
       cat->vp[i].var = v[i];
@@ -188,15 +212,19 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 
   assert (NULL == cat->reverse_variable_map);
 
-  for (i = 0 ; i < cat->n_vars; ++i)
+  for (i = 0 ; i < cat->n_vp; ++i)
     {
       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 = value_hash (val, width, 0);
+      size_t hash;
+      struct value_node *node ;
 
-      struct value_node  *node = lookup_value (&cat->vp[i].map, var, val);
+      if ( var_is_value_missing (var, val, cat->exclude))
+       continue;
 
+      hash = value_hash (val, width, 0);
+      node = lookup_value (&cat->vp[i].map, var, val);
 
       if ( NULL == node)
        {
@@ -207,7 +235,11 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
          node->cc = 0.0;
 
          hmap_insert (&cat->vp[i].map, &node->node,  hash);
-         cat->n_cats_total ++;
+         cat->n_cats_total++;
+         
+         if ( 0 == cat->vp[i].n_cats)
+           cat->n_vars++;
+
          node->subscript = cat->vp[i].n_cats++ ;
        }
 
@@ -224,19 +256,6 @@ categoricals_n_count (const struct categoricals *cat, size_t n)
 }
 
 
-/* Return the index for value VAL in the Nth variable */
-int
-categoricals_index (const struct categoricals *cat, size_t n, const union value *val)
-{
-  struct value_node *vn = lookup_value (&cat->vp[n].map, cat->vp[n].var, val);
-
-  if ( vn == NULL)
-    return -1;
-
-  return vn->subscript;
-}
-
-
 /* Return the total number of categories */
 size_t
 categoricals_total (const struct categoricals *cat)
@@ -258,9 +277,11 @@ categoricals_done (struct categoricals *cat)
   */
   int v;
   int idx = 0;
-  cat->reverse_variable_map = pool_calloc (cat->pool, cat->n_cats_total, sizeof *cat->reverse_variable_map);
+  cat->reverse_variable_map = pool_calloc (cat->pool,
+                                          cat->n_cats_total - cat->n_vars,
+                                          sizeof *cat->reverse_variable_map);
   
-  for (v = 0 ; v < cat->n_vars; ++v)
+  for (v = 0 ; v < cat->n_vp; ++v)
     {
       int i;
       struct var_params *vp = &cat->vp[v];
@@ -277,9 +298,14 @@ categoricals_done (struct categoricals *cat)
          vp->reverse_value_map[vn->subscript] = vn;
        }
 
-      for (i = 0; i < vp->n_cats; ++i)
+      /* Populate the reverse variable map.
+        This implementation considers the first value of each categorical variable
+        as the basis.  Therefore, this loop starts from 1 instead of 0 */
+      for (i = 1; i < vp->n_cats; ++i)
        cat->reverse_variable_map[idx++] = v;
     }
+
+  assert (cat->n_vars <= cat->n_vp);
 }
 
 
@@ -310,6 +336,26 @@ categoricals_get_value_by_subscript (const struct categoricals *cat, int subscri
 }
 
 
+double
+categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscript)
+{
+  int vindex = cat->reverse_variable_map[subscript];
+  const struct var_params *vp = &cat->vp[vindex];
+
+  return vp->cc;
+}
+
+double
+categoricals_get_sum_by_subscript (const struct categoricals *cat, int subscript)
+{
+  int vindex = cat->reverse_variable_map[subscript];
+  const struct var_params *vp = &cat->vp[vindex];
+
+  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript];
+  return vn->cc;
+}
+
+
 /* Returns unity if the value in case C at SUBSCRIPT is equal to the category
    for that subscript */
 double
@@ -323,3 +369,10 @@ categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscr
 
   return value_equal (val, categoricals_get_value_by_subscript (cat, subscript), width);
 }
+
+
+size_t
+categoricals_get_n_variables (const struct categoricals *cat)
+{
+  return cat->n_vars;
+}