X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.c;h=a2cf9dd7854c130afee0e6de6cda28b506e9d486;hb=3b45b998691d080fcb64cb8783633cf7b14a54ff;hp=22d806b347a1d4fd85c725a3b0fe6dd5d3aaa707;hpb=15aae7ade3a0835939d28d78f20ed1eb85500d86;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 22d806b347..a2cf9dd785 100644
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -32,23 +32,26 @@
 
 #include "gl/xalloc.h"
 
+#define EFFECTS_CODING 1
+
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
 
   union value val;            /* The value */
 
-  void *user_data;            /* A pointer to data which the caller can store stuff */
+  int index;                  /* A zero based unique index for this value */
 };
 
 struct interaction_value
 {
-  struct hmap_node node;        /* Node in hash map. */
+  struct hmap_node node;      /* Node in hash map */
+
+  struct ccase *ccase;        /* A case (probably the first in the dataset) which matches this value */
 
-  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 */
 
-  /* Total of the weights of cases matching this interaction */
-  double cc; 
+  void *user_data;            /* A pointer to data which the caller can store stuff */
 };
 
 static struct value_node *
@@ -70,8 +73,10 @@ struct variable_node
   const struct variable *var; /* The variable */
 
   struct hmap valmap;         /* A map of value nodes */
+  int n_vals;                 /* Number of values for this variable */
 };
 
+#if 0
 static void
 dump_interaction (const struct interaction *iact)
 {
@@ -80,6 +85,7 @@ dump_interaction (const struct interaction *iact)
   printf ("Interaction: %s\n", ds_cstr (&str));
   ds_destroy (&str);
 }
+#endif
 
 
 static struct variable_node *
@@ -111,8 +117,12 @@ struct interact_params
   /* The number of distinct values of this interaction */
   int n_cats;
 
-  /* The degrees of freedom for this interaction */
-  int df; 
+  /* 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 preceeding variables */
+  int *df_prod; 
+
+  double *enc_sum;
 
   /* A map of interaction_values indexed by subscript */
   struct interaction_value **reverse_interaction_value_map;
@@ -121,18 +131,17 @@ struct interact_params
 };
 
 
-#if 0
 /* 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_;
+
+  const struct interact_params *iap = aux;
 
-  return interaction_case_cmp_3way (vp->iact, (*vn1)->ccase, (*vn2)->ccase);
+  return interaction_case_cmp_3way (iap->iact, (*vn1p)->ccase, (*vn2p)->ccase);
 }
-#endif
 
 struct categoricals
 {
@@ -180,12 +189,12 @@ struct categoricals
   void *aux2;
 };
 
-
+#if 0
 static void
 categoricals_dump (const struct categoricals *cat)
 {
   int i;
-#if 1
+
   printf ("Reverse Variable Map (short):\n");
   for (i = 0; i < cat->df_sum; ++i)
     {
@@ -193,6 +202,14 @@ categoricals_dump (const struct categoricals *cat)
     }
   printf ("\n");
 
+  printf ("Reverse Variable Map (long):\n");
+  for (i = 0; i < cat->n_cats_total; ++i)
+    {
+      printf (" %d", cat->reverse_variable_map_long[i]);
+    }
+  printf ("\n");
+
+
   printf ("Number of interactions %d\n", cat->n_iap);
   for (i = 0 ; i < cat->n_iap; ++i)
     {
@@ -229,13 +246,40 @@ categoricals_dump (const struct categoricals *cat)
 	}
       printf (")\n");
     }
-#endif
 }
-
+#endif
 
 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)
+	{
+	  case_unref (iv->ccase);
+	}
+
+      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);
+    }
+
+  hmap_destroy (&cat->varmap);
+
+  pool_destroy (cat->pool);
+
   free (cat);
 }
 
@@ -245,7 +289,7 @@ static struct interaction_value *
 lookup_case (const struct hmap *map, const struct interaction *iact, const struct ccase *c)
 {
   struct interaction_value *iv = NULL;
-  size_t hash = interaction_case_hash (iact, c);
+  size_t hash = interaction_case_hash (iact, c, 0);
 
   HMAP_FOR_EACH_WITH_HASH (iv, struct interaction_value, node, hash, map)
     {
@@ -301,6 +345,7 @@ categoricals_create (struct interaction *const*inter, size_t n_inter,
 	    {
 	      vn = pool_malloc (cat->pool, sizeof *vn);
 	      vn->var = var;
+	      vn->n_vals = 0;
 	      hmap_init (&vn->valmap);
 
 	      hmap_insert (&cat->varmap, &vn->node,  hash);
@@ -335,13 +380,13 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
       if (valn == NULL)
 	{
 	  valn = pool_malloc (cat->pool, sizeof *valn);
+	  valn->index = vn->n_vals++;
 	  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;
@@ -349,7 +394,7 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
       //      if ( interaction_case_is_missing (iact, c, cat->exclude))
       //         continue;
 
-      size_t hash = interaction_case_hash (iact, c);
+      size_t hash = interaction_case_hash (iact, c, 0);
       struct interaction_value *node = lookup_case (&cat->iap[i].ivmap, iact, c);
 
       if ( NULL == node)
@@ -361,8 +406,8 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 
 	  hmap_insert (&cat->iap[i].ivmap, &node->node, hash);
 
-	  //	  if (cat->user_data_create)
-	  //       node->user_data = cat->user_data_create (cat->aux1, cat->aux2);
+	  if (cat->user_data_create)
+	    node->user_data = cat->user_data_create (cat->aux1, cat->aux2);
 	}
       else
 	{
@@ -370,9 +415,8 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 	}
       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->update)
+      	cat->update (node->user_data, cat->exclude, cat->wv, NULL, c, cat->aux1, cat->aux2);
     }
 }
 
@@ -387,7 +431,8 @@ categoricals_n_count (const struct categoricals *cat, size_t n)
 size_t
 categoricals_df (const struct categoricals *cat, size_t n)
 {
-  return cat->iap[n].df;
+  const struct interact_params *iap = &cat->iap[n];
+  return iap->df_prod[iap->iact->n_vars - 1];
 }
 
 
@@ -395,6 +440,8 @@ categoricals_df (const struct categoricals *cat, size_t n)
 size_t
 categoricals_n_total (const struct categoricals *cat)
 {
+  assert (cat->reverse_variable_map_long);
+
   return cat->n_cats_total;
 }
 
@@ -426,22 +473,26 @@ categoricals_done (const struct categoricals *cat_)
   /* 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 = 1;
-      cat->iap[i].n_cats = 1;
+      cat->iap[i].df_prod = xcalloc (iact->n_vars, sizeof (int));
 
+      cat->iap[i].n_cats = 1;
+      
       for (v = 0 ; v < iact->n_vars; ++v)
 	{
 	  const struct variable *var = iact->vars[v];
 
 	  struct variable_node *vn = lookup_variable (&cat->varmap, var, hash_pointer (var, 0));
 
-	  cat->iap[i].df *= hmap_count (&vn->valmap) - 1;
+	  cat->iap[i].df_prod[v] = df * (hmap_count (&vn->valmap) - 1);
+	  df = cat->iap[i].df_prod[v];
+
 	  cat->iap[i].n_cats *= hmap_count (&vn->valmap);
 	}
 
-      cat->df_sum += cat->iap[i].df;
+      cat->df_sum += cat->iap[i].df_prod [v - 1];
       cat->n_cats_total += cat->iap[i].n_cats;
     }
 
@@ -472,14 +523,18 @@ categoricals_done (const struct categoricals *cat_)
 	  iap->reverse_interaction_value_map[x++] = ivn;
 	}
 
-#if 0
+      assert (x <= iap->n_cats);
+
       /* For some purposes (eg CONTRASTS in ONEWAY) the values need to be sorted */
-      sort (vp->reverse_interaction_value_map, vp->n_cats, sizeof (const struct interaction_value *),
-	    compare_value_node, vp);
-#endif
+      sort (iap->reverse_interaction_value_map, x, sizeof (*iap->reverse_interaction_value_map),
+	    compare_interaction_value_3way, iap);
+
+      /* Fill the remaining values with null */
+      for (ii = x ; ii < iap->n_cats; ++ii)
+	iap->reverse_interaction_value_map[ii] = NULL;
 
       /* Populate the reverse variable maps. */
-      for (ii = 0; ii < iap->df; ++ii)
+      for (ii = 0; ii < iap->df_prod [iap->iact->n_vars - 1]; ++ii)
 	cat->reverse_variable_map_short[idx_short++] = i;
 
       for (ii = 0; ii < iap->n_cats; ++ii)
@@ -487,6 +542,30 @@ categoricals_done (const struct categoricals *cat_)
     }
 
   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 [iact->n_vars - 1];
+
+      iap->enc_sum = xcalloc (df, sizeof (*(iap->enc_sum)));
+
+      for (y = 0; y < hmap_count (&iap->ivmap); ++y)
+	{
+	  struct interaction_value *iv = iap->reverse_interaction_value_map[y];
+	  for (x = iap->base_subscript_short; x < iap->base_subscript_short + df ;++x)
+	    {
+	      const double bin = categoricals_get_binary_by_subscript (cat, x, iv->ccase); \
+	      iap->enc_sum [x - iap->base_subscript_short] += bin * iv->cc;
+	    }
+	}
+    }
 }
 
 
@@ -503,7 +582,6 @@ reverse_variable_lookup_short (const struct categoricals *cat, int subscript)
 static int
 reverse_variable_lookup_long (const struct categoricals *cat, int subscript)
 {
-  printf ("%s\n", __FUNCTION__);
   assert (cat->reverse_variable_map_long);
   assert (subscript >= 0);
   assert (subscript < cat->n_cats_total);
@@ -521,7 +599,6 @@ categoricals_get_interaction_by_subscript (const struct categoricals *cat, int s
   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)
@@ -529,6 +606,10 @@ categoricals_get_case_by_subscript (const struct categoricals *cat, int subscrip
   int vindex = reverse_variable_lookup_short (cat, subscript);
   const struct interact_params *vp = &cat->iap[vindex];
   const struct interaction_value *vn = vp->reverse_interaction_value_map [subscript - vp->base_subscript_short];
+
+  if ( vn == NULL)
+    return NULL;
+
   return vn->ccase;
 }
 
@@ -548,22 +629,59 @@ categoricals_get_sum_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 interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_short];
-
-  return iv->cc;
+  return   vp->enc_sum[subscript - vp->base_subscript_short];
 }
 
 /* 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,
+categoricals_get_binary_by_subscript (const struct categoricals *cat,
+				      int subscript,
 				      const struct ccase *c)
 {
   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 = reverse_variable_lookup_short (cat, subscript);
+
+  const int base_index = cat->iap[i].base_subscript_short;
+
+  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];
+
+    const union value *val = case_data (c, var);
+    const int width = var_get_width (var);
+    const struct variable_node *vn = lookup_variable (&cat->varmap, var, hash_pointer (var, 0));
+
+    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;
 
-  return interaction_case_equal (iact, c, c2);
+    /* 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
+    if ( valn->index == df )
+      bin = -1.0;
+    else 
+#endif
+    if ( valn->index  != index )
+      bin = 0;
+    
+    result *= bin;
+  }
+
+  return result;
 }
 
 
@@ -574,18 +692,15 @@ categoricals_get_n_variables (const struct categoricals *cat)
   return cat->n_vars;
 }
 
-
-
 /* 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 interact_params *vp = &cat->iap[vindex];
-  //  const struct interaction_value *vn = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
+  const struct interaction_value *vn = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
 
-  //  return vn->ccase;
-  return NULL;
+  return vn->ccase;
 }
 
 void *
@@ -594,8 +709,10 @@ categoricals_get_user_data_by_category (const struct categoricals *cat, int subs
   int vindex = reverse_variable_lookup_long (cat, subscript);
   const struct interact_params *vp = &cat->iap[vindex];
 
-  //const struct value_node *vn = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
-  //return vn->user_data;
-  assert (0);
-  return NULL;
+  const struct interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
+  return iv->user_data;
 }
+
+
+
+