X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;ds=sidebyside;f=src%2Fmath%2Fcategoricals.c;h=8ea2c091f2cab98e16269262d8bacdf56ca5b121;hb=5f3cd72e75cfe4ded7c2ca3b2cd0936412ba71ab;hp=218ac6c180701c78ea424eaa25b6a46fa3f8c1e6;hpb=dc32b2366bbb293a81b57a9a19285188db28c881;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 218ac6c180..8ea2c091f2 100644
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 2009, 2010, 2011 Free Software Foundation, Inc.
+   Copyright (C) 2009, 2010, 2011, 2012, 2014 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -19,6 +19,7 @@
 #include "math/categoricals.h"
 #include "math/interaction.h"
 
+#include <float.h>
 #include <stdio.h>
 
 #include "data/case.h"
@@ -28,53 +29,130 @@
 #include "libpspp/hmap.h"
 #include "libpspp/pool.h"
 #include "libpspp/str.h"
+#include "libpspp/hash-functions.h"
 
 #include "gl/xalloc.h"
 
+#define CATEGORICALS_DEBUG 0
+
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
-  struct ccase *ccase;
-  double cc;                  /* The total of the weights of cases with this value */
+
+  union value val;            /* The value */
+
+  int index;                  /* A zero based unique index for this value */
+};
+
+
+struct interaction_value
+{
+  struct hmap_node node;      /* Node in hash map */
+
+  struct ccase *ccase;        /* A case (probably the first in the dataset) which matches
+				 this value */
+
+  double cc;                  /* Total of the weights of cases matching this interaction */
 
   void *user_data;            /* A pointer to data which the caller can store stuff */
+};
 
-  int subscript;              /* A zero based integer, unique within the variable.
-				 Can be used as an index into an array */
+static struct value_node *
+lookup_value (const struct hmap *map, const union value *val, unsigned int hash, int width)
+{
+  struct value_node *vn = NULL;
+  HMAP_FOR_EACH_WITH_HASH (vn, struct value_node, node, hash, map)
+    {
+      if (value_equal (&vn->val, val, width))
+	break;
+    }
+
+  return vn;
+}
+
+struct variable_node
+{
+  struct hmap_node node;      /* Node in hash map. */
+  const struct variable *var; /* The variable */
+
+  struct hmap valmap;         /* A map of value nodes */
 };
 
-struct interact_params
+
+/* Comparison function to sort value_nodes in ascending order */
+static int
+compare_value_node_3way (const void *vn1_, const void *vn2_, const void *aux)
+{
+  const struct value_node *const *vn1p = vn1_;
+  const struct value_node *const *vn2p = vn2_;
+
+  const struct variable_node *vn = aux;
+
+
+  return value_compare_3way (&(*vn1p)->val, &(*vn2p)->val, var_get_width (vn->var));
+}
+
+
+
+static struct variable_node *
+lookup_variable (const struct hmap *map, const struct variable *var, unsigned int hash)
 {
-  /* A map indexed by a union values */
-  struct hmap map;
+  struct variable_node *vn = NULL;
+  HMAP_FOR_EACH_WITH_HASH (vn, struct variable_node, node, hash, map)
+    {
+      if (vn->var == var)
+	break;
+    }
+
+  return vn;
+}
 
+
+struct interact_params
+{
+  /* The interaction, and an array with iact->n_vars elements such that
+     varnodes[x] points to the variable_node for iact->vars[x]. */
   const struct interaction *iact;
+  struct variable_node **varnodes;
+
+  /* An example of each interaction that appears in the data, like a frequency
+     table for 'iact'.  By construction, the number of elements must be less
+     than or equal to 'n_cats'.
 
-  int base_subscript_short;
-  int base_subscript_long;
+     categoricals_update() updates 'ivmap' case-by-case, then
+     categoricals_done() dumps 'ivmap' into 'ivs' and sorts it. */
+  struct hmap ivmap;
+  struct interaction_value **ivs;
 
-  /* The number of distinct values of this variable */
+  int base_df;
+  int base_cats;
+
+  /* Product of hmap_count(&varnodes[*]->valmap), that is, the maximum number
+     of distinct values of this interaction. */
   int n_cats;
 
-  /* A map of values indexed by subscript */
-  const struct value_node **reverse_value_map;
+  /* An array of integers df_n * df_{n-1} * df_{n-2} ...
+     These are the products of the degrees of freedom for the current
+     variable and all preceding variables */
+  int *df_prod;
+
+  double *enc_sum;
 
-  /* Total of the weights of this variable */
-  double cc; 
+  /* Sum of ivs[*]->cc. */
+  double cc;
 };
 
 
-/* Comparison function to sort the reverse_value_map in ascending order */
 static int
-compare_value_node (const void *vn1_, const void *vn2_, const void *aux)
+compare_interaction_value_3way (const void *vn1_, const void *vn2_, const void *aux)
 {
-  const struct value_node * const *vn1 = vn1_;
-  const struct value_node * const *vn2 = vn2_;
-  const struct interact_params *vp = aux;
+  const struct interaction_value *const *vn1p = vn1_;
+  const struct interaction_value *const *vn2p = vn2_;
 
-  return interaction_case_cmp_3way (vp->iact, (*vn1)->ccase, (*vn2)->ccase);
-}
+  const struct interact_params *iap = aux;
 
+  return interaction_case_cmp_3way (iap->iact, (*vn1p)->ccase, (*vn2p)->ccase);
+}
 
 struct categoricals
 {
@@ -83,182 +161,244 @@ struct categoricals
 
   /* An array of interact_params */
   struct interact_params *iap;
-
-  /* The size of IAP. (ie, the number of interactions involved.) */
   size_t n_iap;
 
+  /* Map whose members are the union of the variables which comprise IAP */
+  struct hmap varmap;
+
   /* The number of categorical variables which contain entries.
      In the absence of missing values, this will be equal to N_IAP */
   size_t n_vars;
 
   /* A map to enable the lookup of variables indexed by subscript.
      This map considers only the N - 1 of the N variables.
-   */
-  int *reverse_variable_map_short;
+  */
+  int *df_to_iact; /* 'df_sum' elements. */
+  size_t df_sum;
 
   /* Like the above, but uses all N variables */
-  int *reverse_variable_map_long;
-
+  int *cat_to_iact; /* 'n_cats_total' elements. */
   size_t n_cats_total;
 
   struct pool *pool;
 
-  /* Missing values to be excluded */
-  enum mv_class exclude;
+  /* Missing values in the factor variables to be excluded */
+  enum mv_class fctr_excl;
 
-  /* Function to be called on each update */
-  update_func *update;
+  const void *aux1;
+  void *aux2;
 
-  /* Function specified by the caller to create user_data */
-  user_data_create_func *user_data_create;
+  bool sane;
 
-  /* Auxilliary data to be passed to update and user_data_create_func*/
-  void *aux1;
-  void *aux2;
+  const struct payload *payload;
 };
 
 
-void
-categoricals_destroy ( struct categoricals *cat)
+bool
+categoricals_isbalanced (const struct categoricals *cat)
 {
   int i;
-  if (cat != NULL)
+
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      for (i = 0 ; i < cat->n_iap; ++i)
-	{
-	  struct hmap *map = &cat->iap[i].map;
-	  struct value_node *nn;
+      int v;
+      const struct interact_params *iap = &cat->iap[i];
 
-	  HMAP_FOR_EACH (nn, struct value_node, node, map)
-	    {
-	      case_unref (nn->ccase);
-	    }	  
-	  
-	  hmap_destroy (map);
+      double oval = -1.0;
+      for (v = 0; v < hmap_count (&iap->ivmap); ++v)
+	{
+	  const struct interaction_value *iv = iap->ivs[v];
+	  if (oval == -1.0)
+	    oval = iv->cc;
+	  if (oval != iv->cc)
+	    return false;
 	}
-      
-      pool_destroy (cat->pool);
-      free (cat);
     }
+  return true;
 }
 
 
-#if 0
-void
+static void
 categoricals_dump (const struct categoricals *cat)
 {
-  int v;
-
-  for (v = 0 ; v < cat->n_iap; ++v)
+  if (CATEGORICALS_DEBUG)
     {
-      const struct interact_params *vp = &cat->iap[v];
-      const struct hmap *m = &vp->map;
-      struct hmap_node *node ;
-      int x;
-      struct string str ;
-      ds_init_empty (&str);
-
-      interaction_to_string (vp->iact, &str);
-      printf ("\n%s (%d)  CC=%g n_cats=%d:\n", 
-	      ds_cstr (&str),
-	      vp->base_subscript_long, vp->cc, vp->n_cats);
-
-#if 0
-      printf ("Reverse map\n");
-      for (x = 0 ; x < vp->n_cats; ++x)
+      int i;
+
+      printf ("df to interaction map:\n");
+      for (i = 0; i < cat->df_sum; ++i)
 	{
-	  struct string s;
-	  const struct value_node *vn = vp->reverse_value_map[x];
-	  ds_init_empty (&s);
-	  var_append_value_name (vp->var, &vn->value, &s);
-	  printf ("Value for %d is %s\n", x, ds_cstr(&s));
-	  ds_destroy (&s);
+	  printf (" %d", cat->df_to_iact[i]);
 	}
+      printf ("\n");
 
-      printf ("\nForward map\n");
-      for (node = hmap_first (m); node; node = hmap_next (m, node))
+      printf ("Category to interaction map:\n");
+      for (i = 0; i < cat->n_cats_total; ++i)
 	{
-	  struct string s;
-	  const struct value_node *vn = HMAP_DATA (node, struct value_node, node);
-	  ds_init_empty (&s);
-	  var_append_value_name (vp->var, &vn->value, &s);
-	  printf ("Value: %s; Index %d; CC %g\n",
-		  ds_cstr (&s),
-		  vn->subscript, vn->cc);
-	  ds_destroy (&s);
+	  printf (" %d", cat->cat_to_iact[i]);
+	}
+      printf ("\n");
+
+      printf ("Number of interactions %zu\n", cat->n_iap);
+      for (i = 0 ; i < cat->n_iap; ++i)
+	{
+	  int v;
+	  struct string str;
+	  const struct interact_params *iap = &cat->iap[i];
+	  const struct interaction *iact = iap->iact;
+
+	  ds_init_empty (&str);
+	  interaction_to_string (iact, &str);
+
+	  printf ("\nInteraction: \"%s\" (number of categories: %d); ", ds_cstr (&str), iap->n_cats);
+	  ds_destroy (&str);
+	  printf ("Base index (df/categories): %d/%d\n", iap->base_df, iap->base_cats);
+
+	  printf ("\t(");
+	  for (v = 0; v < hmap_count (&iap->ivmap); ++v)
+	    {
+	      int vv;
+	      const struct interaction_value *iv = iap->ivs[v];
+
+	      if (v > 0)  printf ("   ");
+	      printf ("{");
+	      for (vv = 0; vv < iact->n_vars; ++vv)
+		{
+		  const struct variable *var = iact->vars[vv];
+		  const union value *val = case_data (iv->ccase, var);
+		  struct variable_node *vn = iap->varnodes[vv];
+
+		  const int width = var_get_width (var);
+		  unsigned int valhash = value_hash (val, width, 0);
+		  struct value_node *valn = lookup_value (&vn->valmap, val, valhash, width);
+
+		  assert (vn->var == var);
+
+		  printf ("%.*g(%d)", DBL_DIG + 1, val->f, valn->index);
+		  if (vv < iact->n_vars - 1)
+		    printf (", ");
+		}
+	      printf ("}");
+	    }
+	  printf (")\n");
 	}
-#endif
     }
+}
 
-  assert (cat->n_vars <= cat->n_iap);
+void
+categoricals_destroy (struct categoricals *cat)
+{
+  struct variable_node *vn = NULL;
+  int i;
+  if (NULL == cat)
+    return;
+
+  for (i = 0; i < cat->n_iap; ++i)
+    {
+      struct interaction_value *iv = NULL;
+      /* Interate over each interaction value, and unref any cases that we reffed */
+      HMAP_FOR_EACH (iv, struct interaction_value, node, &cat->iap[i].ivmap)
+	{
+	  if (cat->payload && cat->payload->destroy)
+	    cat->payload->destroy (cat->aux1, cat->aux2, iv->user_data);
+	  case_unref (iv->ccase);
+	}
 
-  printf ("\n");
-  printf ("Number of interactions: %d\n", cat->n_iap);
-  printf ("Number of non-empty categorical variables: %d\n", cat->n_vars);
-  printf ("Total number of categories: %d\n", cat->n_cats_total);
+      free (cat->iap[i].enc_sum);
+      free (cat->iap[i].df_prod);
+      hmap_destroy (&cat->iap[i].ivmap);
+    }
+
+  /* Interate over each variable and delete its value map */
+  HMAP_FOR_EACH (vn, struct variable_node, node, &cat->varmap)
+    {
+      hmap_destroy (&vn->valmap);
+    }
 
-  printf ("\nReverse variable map (short):\n");
-  for (v = 0 ; v < cat->n_cats_total - cat->n_vars; ++v)
-    printf ("%d ", cat->reverse_variable_map_short[v]);
+  hmap_destroy (&cat->varmap);
 
-  printf ("\nReverse variable map (long):\n");
-  for (v = 0 ; v < cat->n_cats_total; ++v)
-    printf ("%d ", cat->reverse_variable_map_long[v]);
+  pool_destroy (cat->pool);
 
-  printf ("\n");
+  free (cat);
 }
-#endif
 
 
-static struct value_node *
+
+static struct interaction_value *
 lookup_case (const struct hmap *map, const struct interaction *iact, const struct ccase *c)
 {
-  struct value_node *nn;
-  size_t hash = interaction_case_hash (iact, c);
+  struct interaction_value *iv = NULL;
+  size_t hash = interaction_case_hash (iact, c, 0);
 
-  HMAP_FOR_EACH_WITH_HASH (nn, struct value_node, node, hash, map)
+  HMAP_FOR_EACH_WITH_HASH (iv, struct interaction_value, node, hash, map)
     {
-      if (interaction_case_equal (iact, c, nn->ccase))
+      if (interaction_case_equal (iact, c, iv->ccase))
 	break;
-
-      fprintf (stderr, "Warning: Hash table collision\n");
     }
 
-  return nn;
+  return iv;
 }
 
+/* Returns true iff CAT is sane, that is, if it is complete and has at least
+   one value. */
+bool
+categoricals_sane (const struct categoricals *cat)
+{
+  return cat->sane;
+}
 
+/* Creates and returns a new categoricals object whose variables come from the
+   N_INTER interactions objects in the array starting at INTER.  (The INTER
+   objects must outlive the categoricals object because it uses them
+   internally.)
+
+   FCTR_EXCL determines which cases are listwise ignored by
+   categoricals_update(). */
 struct categoricals *
-categoricals_create (const struct interaction **inter, size_t n_inter,
-		     const struct variable *wv, enum mv_class exclude,
-		     user_data_create_func *udf,
-		     update_func *update, void *aux1, void *aux2
-		     )
+categoricals_create (struct interaction *const*inter, size_t n_inter,
+		     const struct variable *wv, enum mv_class fctr_excl)
 {
   size_t i;
   struct categoricals *cat = xmalloc (sizeof *cat);
-  
+
   cat->n_iap = n_inter;
   cat->wv = wv;
   cat->n_cats_total = 0;
   cat->n_vars = 0;
-  cat->reverse_variable_map_short = NULL;
-  cat->reverse_variable_map_long = NULL;
+  cat->df_to_iact = NULL;
+  cat->cat_to_iact = NULL;
   cat->pool = pool_create ();
-  cat->exclude = exclude;
-  cat->update = update;
-  cat->user_data_create = udf;
-
-  cat->aux1 = aux1;
-  cat->aux2 = aux2;
-
+  cat->fctr_excl = fctr_excl;
+  cat->payload = NULL;
+  cat->aux2 = NULL;
+  cat->sane = false;
 
   cat->iap = pool_calloc (cat->pool, cat->n_iap, sizeof *cat->iap);
 
+  hmap_init (&cat->varmap);
   for (i = 0 ; i < cat->n_iap; ++i)
     {
-      hmap_init (&cat->iap[i].map);
+      int v;
+      hmap_init (&cat->iap[i].ivmap);
       cat->iap[i].iact = inter[i];
+      cat->iap[i].cc = 0.0;
+      cat->iap[i].varnodes = pool_nmalloc (cat->pool, cat->iap[i].iact->n_vars,
+                                           sizeof *cat->iap[i].varnodes);
+      for (v = 0; v < inter[i]->n_vars; ++v)
+	{
+	  const struct variable *var = inter[i]->vars[v];
+	  unsigned int hash = hash_pointer (var, 0);
+	  struct variable_node *vn = lookup_variable (&cat->varmap, var, hash);
+	  if (vn == NULL)
+	    {
+	      vn = pool_malloc (cat->pool, sizeof *vn);
+	      vn->var = var;
+	      hmap_init (&vn->valmap);
+
+	      hmap_insert (&cat->varmap, &vn->node,  hash);
+	    }
+          cat->iap[i].varnodes[v] = vn;
+	}
     }
 
   return cat;
@@ -269,70 +409,127 @@ categoricals_create (const struct interaction **inter, size_t n_inter,
 void
 categoricals_update (struct categoricals *cat, const struct ccase *c)
 {
-  size_t i;
-  
-  const double weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
+  int i;
+  struct variable_node *vn = NULL;
+  double weight;
+
+  if (NULL == cat)
+    return;
 
-  assert (NULL == cat->reverse_variable_map_short);
-  assert (NULL == cat->reverse_variable_map_long);
+  weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
+  weight = var_force_valid_weight (cat->wv, weight, NULL);
+
+  assert (NULL == cat->df_to_iact);
+  assert (NULL == cat->cat_to_iact);
+
+  /* Interate over each variable, and add the value of that variable
+     to the appropriate map, if it's not already present. */
+  HMAP_FOR_EACH (vn, struct variable_node, node, &cat->varmap)
+    {
+      const int width = var_get_width (vn->var);
+      const union value *val = case_data (c, vn->var);
+      unsigned int hash = value_hash (val, width, 0);
+
+      struct value_node *valn = lookup_value (&vn->valmap, val, hash, width);
+      if (valn == NULL)
+	{
+	  valn = pool_malloc (cat->pool, sizeof *valn);
+	  valn->index = -1;
+	  value_init (&valn->val, width);
+	  value_copy (&valn->val, val, width);
+	  hmap_insert (&vn->valmap, &valn->node, hash);
+	}
+    }
 
   for (i = 0 ; i < cat->n_iap; ++i)
     {
       const struct interaction *iact = cat->iap[i].iact;
+
       size_t hash;
-      struct value_node *node ;
+      struct interaction_value *node;
 
-      if ( interaction_case_is_missing (iact, c, cat->exclude))
+      if ( interaction_case_is_missing (iact, c, cat->fctr_excl))
 	continue;
 
-      hash = interaction_case_hash (iact, c);
-      node = lookup_case (&cat->iap[i].map, iact, c);
+      hash = interaction_case_hash (iact, c, 0);
+      node = lookup_case (&cat->iap[i].ivmap, iact, c);
 
       if ( NULL == node)
 	{
 	  node = pool_malloc (cat->pool, sizeof *node);
-
 	  node->ccase = case_ref (c);
-	  node->cc = 0.0;
+	  node->cc = weight;
 
-	  hmap_insert (&cat->iap[i].map, &node->node,  hash);
-	  cat->n_cats_total++;
-	  
-	  if ( 0 == cat->iap[i].n_cats)
-	    cat->n_vars++;
+	  hmap_insert (&cat->iap[i].ivmap, &node->node, hash);
 
-	  node->subscript = cat->iap[i].n_cats++ ;
-
-	  if (cat->user_data_create)
-	    node->user_data = cat->user_data_create (cat->aux1, cat->aux2);
+	  if (cat->payload)
+	    {
+	      node->user_data = cat->payload->create (cat->aux1, cat->aux2);
+	    }
+	}
+      else
+	{
+	  node->cc += weight;
 	}
-
-      node->cc += weight;
       cat->iap[i].cc += weight;
 
-      if (cat->update)
-	cat->update (node->user_data, cat->exclude, cat->wv, NULL, c, cat->aux1, cat->aux2);
+      if (cat->payload)
+	{
+	  cat->payload->update (cat->aux1, cat->aux2, node->user_data, c, weight);
+	}
     }
 }
 
-/* Return the number of categories (distinct values) for variable N */
+/* Return the number of categories (distinct values) for interaction IDX in
+   CAT. */
 size_t
 categoricals_n_count (const struct categoricals *cat, size_t n)
 {
-  return hmap_count (&cat->iap[n].map);
+  return hmap_count (&cat->iap[n].ivmap);
+}
+
+
+/* Returns the number of degrees of freedom for interaction IDX within CAT. */
+size_t
+categoricals_df (const struct categoricals *cat, size_t n)
+{
+  const struct interact_params *iap = &cat->iap[n];
+  return iap->df_prod[iap->iact->n_vars - 1];
 }
 
 
-/* Return the total number of categories */
+/* Return the total number of categories across all interactions in CAT. */
 size_t
-categoricals_total (const struct categoricals *cat)
+categoricals_n_total (const struct categoricals *cat)
 {
+  if (!categoricals_is_complete (cat))
+    return 0;
+
   return cat->n_cats_total;
 }
 
+/* Returns the total degrees of freedom for CAT. */
+size_t
+categoricals_df_total (const struct categoricals *cat)
+{
+  if (NULL == cat)
+    return 0;
+
+  return cat->df_sum;
+}
+
+/* Returns true iff categoricals_done() has been called for CAT. */
+bool
+categoricals_is_complete (const struct categoricals *cat)
+{
+  return (NULL != cat->df_to_iact);
+}
 
-/* This function must be called *before* any call to categoricals_get_*_by subscript and
- *after* all calls to categoricals_update */
+
+/* This function must be called (once) before any call to the *_by_subscript or
+  *_by_category functions, but AFTER any calls to categoricals_update.  If this
+  function returns false, then no calls to _by_subscript or *_by_category are
+  allowed. */
 void
 categoricals_done (const struct categoricals *cat_)
 {
@@ -344,68 +541,160 @@ categoricals_done (const struct categoricals *cat_)
   */
   struct categoricals *cat = CONST_CAST (struct categoricals *, cat_);
   int v;
-  int idx_short = 0;
-  int idx_long = 0;
-  cat->reverse_variable_map_short = pool_calloc (cat->pool,
-						 cat->n_cats_total - cat->n_vars,
-						 sizeof *cat->reverse_variable_map_short);
-
-  cat->reverse_variable_map_long = pool_calloc (cat->pool,
-						cat->n_cats_total,
-						sizeof *cat->reverse_variable_map_long);
-  
-  for (v = 0 ; v < cat->n_iap; ++v)
+  int i;
+  int idx_df = 0;
+  int idx_cat = 0;
+
+  if (NULL == cat)
+    return;
+
+  cat->df_sum = 0;
+  cat->n_cats_total = 0;
+
+  /* Assign 'index' to each variables' value_nodes, counting up from 0 in
+     ascending order by value. */
+  struct variable_node *vn;
+  HMAP_FOR_EACH (vn, struct variable_node, node, &cat->varmap)
     {
-      int i;
-      struct interact_params *vp = &cat->iap[v];
-      int n_cats_total = categoricals_n_count (cat, v);
-      struct hmap_node *node ;
+      size_t n_vals = hmap_count (&vn->valmap);
+      if (!n_vals)
+        {
+          cat->sane = false;
+          return;
+        }
+
+      struct value_node **nodes = xcalloc (sizeof *nodes, n_vals);
+      int x = 0;
+      struct value_node *valnd;
+      HMAP_FOR_EACH (valnd, struct value_node, node, &vn->valmap)
+        nodes[x++] = valnd;
+      sort (nodes, n_vals, sizeof *nodes, compare_value_node_3way, vn);
+      for (x = 0; x < n_vals; ++x)
+        nodes[x]->index = x;
+      free (nodes);
+    }
+
+  /* Calculate the degrees of freedom, and the number of categories */
+  for (i = 0 ; i < cat->n_iap; ++i)
+    {
+      int df = 1;
+      const struct interaction *iact = cat->iap[i].iact;
+
+      cat->iap[i].df_prod = iact->n_vars ? xcalloc (iact->n_vars, sizeof (int)) : NULL;
+
+      cat->iap[i].n_cats = 1;
 
-      vp->reverse_value_map = pool_calloc (cat->pool, n_cats_total, sizeof *vp->reverse_value_map);
+      for (v = 0 ; v < iact->n_vars; ++v)
+	{
+	  struct variable_node *vn = cat->iap[i].varnodes[v];
+          size_t n_vals = hmap_count (&vn->valmap);
+	  cat->iap[i].df_prod[v] = df * (n_vals - 1);
+      	  df = cat->iap[i].df_prod[v];
+
+	  cat->iap[i].n_cats *= n_vals;
+	}
+
+      if (v > 0)
+	cat->df_sum += cat->iap[i].df_prod [v - 1];
+
+      cat->n_cats_total += cat->iap[i].n_cats;
+    }
+
+
+  cat->df_to_iact = pool_calloc (cat->pool, cat->df_sum,
+                                 sizeof *cat->df_to_iact);
 
-      vp->base_subscript_short = idx_short;
-      vp->base_subscript_long = idx_long;
+  cat->cat_to_iact = pool_calloc (cat->pool, cat->n_cats_total,
+                                  sizeof *cat->cat_to_iact);
 
-      for (node = hmap_first (&vp->map); node; node = hmap_next (&vp->map, node))
+  for (i = 0 ; i < cat->n_iap; ++i)
+    {
+      struct interaction_value *ivn = NULL;
+      int x = 0;
+      int ii;
+      struct interact_params *iap = &cat->iap[i];
+
+      iap->base_df = idx_df;
+      iap->base_cats = idx_cat;
+
+      iap->ivs = pool_calloc (cat->pool, iap->n_cats, sizeof *iap->ivs);
+
+      HMAP_FOR_EACH (ivn, struct interaction_value, node, &iap->ivmap)
 	{
-	  const struct value_node *vn = HMAP_DATA (node, struct value_node, node);
-	  vp->reverse_value_map[vn->subscript] = vn;
+	  iap->ivs[x++] = ivn;
 	}
 
+      assert (x <= iap->n_cats);
+
       /* For some purposes (eg CONTRASTS in ONEWAY) the values need to be sorted */
-      sort (vp->reverse_value_map, vp->n_cats, sizeof (const struct value_node *),
-	    compare_value_node, vp);
+      sort (iap->ivs, x, sizeof *iap->ivs,
+            compare_interaction_value_3way, iap);
 
-      /* Populate the reverse variable maps. */
-      for (i = 0; i < vp->n_cats - 1; ++i)
-	cat->reverse_variable_map_short[idx_short++] = v;
+      /* Fill the remaining values with null */
+      for (ii = x ; ii < iap->n_cats; ++ii)
+	iap->ivs[ii] = NULL;
 
-      for (i = 0; i < vp->n_cats; ++i)
-	cat->reverse_variable_map_long[idx_long++] = v;
+      /* Populate the variable maps. */
+      if (iap->df_prod)
+	{
+	  for (ii = 0; ii < iap->df_prod [iap->iact->n_vars - 1]; ++ii)
+	    cat->df_to_iact[idx_df++] = i;
+	}
+
+      for (ii = 0; ii < iap->n_cats; ++ii)
+	cat->cat_to_iact[idx_cat++] = i;
     }
 
   assert (cat->n_vars <= cat->n_iap);
+
+  categoricals_dump (cat);
+
+  /* Tally up the sums for all the encodings */
+  for (i = 0 ; i < cat->n_iap; ++i)
+    {
+      int x, y;
+      struct interact_params *iap = &cat->iap[i];
+      const struct interaction *iact = iap->iact;
+
+      const int df = iap->df_prod ? iap->df_prod [iact->n_vars - 1] : 0;
+
+      iap->enc_sum = xcalloc (df, sizeof (*(iap->enc_sum)));
+
+      for (y = 0; y < hmap_count (&iap->ivmap); ++y)
+	{
+	  struct interaction_value *iv = iap->ivs[y];
+	  for (x = iap->base_df; x < iap->base_df + df ;++x)
+	    {
+	      const double bin = categoricals_get_effects_code_for_case (cat, x, iv->ccase);
+	      iap->enc_sum [x - iap->base_df] += bin * iv->cc;
+	    }
+	  if (cat->payload && cat->payload->calculate)
+	    cat->payload->calculate (cat->aux1, cat->aux2, iv->user_data);
+	}
+    }
+
+  cat->sane = true;
 }
 
 
 static int
-reverse_variable_lookup_short (const struct categoricals *cat, int subscript)
+df_to_iap (const struct categoricals *cat, int subscript)
 {
-  assert (cat->reverse_variable_map_short);
+  assert (cat->df_to_iact);
   assert (subscript >= 0);
-  assert (subscript < cat->n_cats_total - cat->n_vars);
+  assert (subscript < cat->df_sum);
 
-  return cat->reverse_variable_map_short[subscript];
+  return cat->df_to_iact[subscript];
 }
 
 static int
-reverse_variable_lookup_long (const struct categoricals *cat, int subscript)
+cat_index_to_iap (const struct categoricals *cat, int subscript)
 {
-  assert (cat->reverse_variable_map_long);
+  assert (cat->cat_to_iact);
   assert (subscript >= 0);
   assert (subscript < cat->n_cats_total);
 
-  return cat->reverse_variable_map_long[subscript];
+  return cat->cat_to_iact[subscript];
 }
 
 
@@ -413,28 +702,15 @@ reverse_variable_lookup_long (const struct categoricals *cat, int subscript)
 const struct interaction *
 categoricals_get_interaction_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int index = reverse_variable_lookup_short (cat, subscript);
+  int index = df_to_iap (cat, subscript);
 
   return cat->iap[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 interact_params *vp = &cat->iap[vindex];
-  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_short];
-
-  return vn->ccase;
-}
-
-
 double
 categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int vindex = reverse_variable_lookup_short (cat, subscript);
+  int vindex = df_to_iap (cat, subscript);
   const struct interact_params *vp = &cat->iap[vindex];
 
   return vp->cc;
@@ -443,53 +719,147 @@ categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscr
 double
 categoricals_get_sum_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int vindex = reverse_variable_lookup_short (cat, subscript);
+  int vindex = df_to_iap (cat, subscript);
   const struct interact_params *vp = &cat->iap[vindex];
 
-  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_short];
-  return vn->cc;
+  return   vp->enc_sum[subscript - vp->base_df];
 }
 
 
 /* Returns unity if the value in case C at SUBSCRIPT is equal to the category
    for that subscript */
-double
-categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscript,
-				      const struct ccase *c)
+static double
+categoricals_get_code_for_case (const struct categoricals *cat, int subscript,
+				const struct ccase *c,
+				bool effects_coding)
 {
   const struct interaction *iact = categoricals_get_interaction_by_subscript (cat, subscript);
 
-  const struct ccase *c2 =  categoricals_get_case_by_subscript (cat, subscript);
+  const int i = df_to_iap (cat, subscript);
+
+  const int base_index = cat->iap[i].base_df;
+
+  int v;
+  double result = 1.0;
+
+  const struct interact_params *iap = &cat->iap[i];
+
+  double dfp = 1.0;
+  for (v = 0; v < iact->n_vars; ++v)
+    {
+      const struct variable *var = iact->vars[v];
 
-  return interaction_case_equal (iact, c, c2);
+      const union value *val = case_data (c, var);
+      const int width = var_get_width (var);
+      const struct variable_node *vn = iap->varnodes[v];
+
+      const unsigned int hash = value_hash (val, width, 0);
+      const struct value_node *valn = lookup_value (&vn->valmap, val, hash, width);
+
+      double bin = 1.0;
+
+      const double df = iap->df_prod[v] / dfp;
+
+      /* Translate the subscript into an index for the individual variable */
+      const int index = ((subscript - base_index) % iap->df_prod[v] ) / dfp;
+      dfp = iap->df_prod [v];
+
+      if (effects_coding && valn->index == df )
+	bin = -1.0;
+      else if ( valn->index != index )
+	bin = 0;
+
+      result *= bin;
+    }
+
+  return result;
 }
 
 
-size_t
-categoricals_get_n_variables (const struct categoricals *cat)
+/* Returns unity if the value in case C at SUBSCRIPT is equal to the category
+   for that subscript */
+double
+categoricals_get_dummy_code_for_case (const struct categoricals *cat, int subscript,
+				     const struct ccase *c)
+{
+  return categoricals_get_code_for_case (cat, subscript, c, false);
+}
+
+/* Returns unity if the value in case C at SUBSCRIPT is equal to the category
+   for that subscript.
+   Else if it is the last category, return -1.
+   Otherwise return 0.
+ */
+double
+categoricals_get_effects_code_for_case (const struct categoricals *cat, int subscript,
+					const struct ccase *c)
 {
-  return cat->n_vars;
+  return categoricals_get_code_for_case (cat, subscript, c, true);
 }
 
+/* Return a case containing the set of values corresponding to
+   the Nth Category of the IACTth interaction */
+const struct ccase *
+categoricals_get_case_by_category_real (const struct categoricals *cat, int iact, int n)
+{
+  const struct interaction_value *vn;
+
+  const struct interact_params *vp = &cat->iap[iact];
+
+  if ( n >= hmap_count (&vp->ivmap))
+    return NULL;
+
+  vn = vp->ivs [n];
+
+  return vn->ccase;
+}
+
+/* Return a the user data corresponding to the Nth Category of the IACTth interaction. */
+void *
+categoricals_get_user_data_by_category_real (const struct categoricals *cat, int iact, int n)
+{
+  const struct interact_params *vp = &cat->iap[iact];
+  const struct interaction_value *iv ;
+
+  if ( n >= hmap_count (&vp->ivmap))
+    return NULL;
+
+  iv = vp->ivs [n];
+
+  return iv->user_data;
+}
+
+
 
 /* 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);
+  int vindex = cat_index_to_iap (cat, subscript);
   const struct interact_params *vp = &cat->iap[vindex];
-  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_long];
+  const struct interaction_value *vn = vp->ivs [subscript - vp->base_cats];
 
   return vn->ccase;
 }
 
-
 void *
 categoricals_get_user_data_by_category (const struct categoricals *cat, int subscript)
 {
-  int vindex = reverse_variable_lookup_long (cat, subscript);
+  int vindex = cat_index_to_iap (cat, subscript);
   const struct interact_params *vp = &cat->iap[vindex];
 
-  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_long];
-  return vn->user_data;
+  const struct interaction_value *iv = vp->ivs [subscript - vp->base_cats];
+  return iv->user_data;
+}
+
+
+
+
+void
+categoricals_set_payload (struct categoricals *cat, const struct payload *p,
+			  const void *aux1, void *aux2)
+{
+  cat->payload = p;
+  cat->aux1 = aux1;
+  cat->aux2 = aux2;
 }