X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.c;h=d3aa372867249807618885cf697d66984a8233fc;hb=32ee0e0402d6d56674f53a47d879ec5c07dabe09;hp=73435175a9af4bede79f1354a60c4442a903e1f8;hpb=40b3d65ffb7a812d1c27e965d87378b0dbf5359b;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 73435175a9..d3aa372867 100644
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -28,53 +28,111 @@
 #include "libpspp/hmap.h"
 #include "libpspp/pool.h"
 #include "libpspp/str.h"
+#include "libpspp/hash-functions.h"
 
 #include "gl/xalloc.h"
 
 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 */
+
+};
+
+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 */
+
+  /* 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 *
+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;
+}
 
-  int subscript;              /* A zero based integer, unique within the variable.
-				 Can be used as an index into an array */
+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 */
 };
 
+static void
+dump_interaction (const struct interaction *iact)
+{
+  struct string str = DS_EMPTY_INITIALIZER;
+  interaction_to_string (iact, &str);
+  printf ("Interaction: %s\n", ds_cstr (&str));
+  ds_destroy (&str);
+}
+
+
+static struct variable_node *
+lookup_variable (const struct hmap *map, const struct variable *var, unsigned int hash)
+{
+  struct variable_node *vn = NULL;
+  HMAP_FOR_EACH_WITH_HASH (vn, struct variable_node, node, hash, map)
+    {
+      if (vn->var == var)
+	break;
+      
+      fprintf (stderr, "Warning: Hash table collision\n");
+    }
+  
+  return vn;
+}
+
+
 struct interact_params
 {
-  /* A map indexed by a union values */
-  struct hmap map;
+  /* A map indexed by a interaction_value */
+  struct hmap ivmap;
 
   const struct interaction *iact;
 
   int base_subscript_short;
   int base_subscript_long;
 
-  /* The number of distinct values of this variable */
+  /* The number of distinct values of this interaction */
   int n_cats;
 
-  /* A map of values indexed by subscript */
-  const struct value_node **reverse_value_map;
+  /* The degrees of freedom for this interaction */
+  int df; 
 
-  /* Total of the weights of this variable */
-  double cc; 
+  /* A map of interaction_values indexed by subscript */
+  struct interaction_value **reverse_interaction_value_map;
+
+  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
 {
@@ -84,6 +142,9 @@ struct categoricals
   /* An array of interact_params */
   struct interact_params *iap;
 
+  /* Map whose members are the union of the variables which comprise IAP */
+  struct hmap varmap;
+
   /* The size of IAP. (ie, the number of interactions involved.) */
   size_t n_iap;
 
@@ -91,6 +152,8 @@ struct categoricals
      In the absence of missing values, this will be equal to N_IAP */
   size_t n_vars;
 
+  size_t df_sum;
+
   /* A map to enable the lookup of variables indexed by subscript.
      This map considers only the N - 1 of the N variables.
    */
@@ -118,112 +181,118 @@ struct categoricals
 };
 
 
-void
-categoricals_destroy ( struct categoricals *cat)
+static void
+categoricals_dump (const struct categoricals *cat)
 {
   int i;
-  if (cat != NULL)
+
+  printf ("Reverse Variable Map (short):\n");
+  for (i = 0; i < cat->df_sum; ++i)
     {
-      for (i = 0 ; i < cat->n_iap; ++i)
-	{
-	  struct hmap *map = &cat->iap[i].map;
-	  struct value_node *nn;
+      printf (" %d", cat->reverse_variable_map_short[i]);
+    }
+  printf ("\n");
 
-	  HMAP_FOR_EACH (nn, struct value_node, node, map)
-	    {
-	      case_unref (nn->ccase);
-	    }	  
+  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)
+    {
+      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 (n: %d; df: %d ); ", ds_cstr (&str), iap->n_cats, iap->df);
+      ds_destroy (&str);
+      printf ("Base subscript: %d\n", iap->base_subscript_short);
+
+      printf ("\t(");
+      for (v = 0; v < hmap_count (&iap->ivmap); ++v)
+	{
+	  int vv;
+	  const struct interaction_value *iv = iap->reverse_interaction_value_map[v];
 	  
-	  hmap_destroy (map);
+	  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);
+	      
+	      printf ("%g", val->f);
+	      if (vv < iact->n_vars - 1)
+		printf (", ");
+	    }
+	  printf ("}");
 	}
-      
-      pool_destroy (cat->pool);
-      free (cat);
+      printf (")\n");
     }
 }
 
 
-#if 0
 void
-categoricals_dump (const struct categoricals *cat)
+categoricals_destroy (struct categoricals *cat)
 {
-  int v;
-
-  for (v = 0 ; v < cat->n_iap; ++v)
+  struct variable_node *vn = NULL;
+  int i;
+  if (NULL == cat)
+    return;
+  for (i = 0; i < cat->n_iap; ++i)
     {
-      const struct interact_params *vp = &cat->vp[v];
-      const struct hmap *m = &vp->map;
-      struct hmap_node *node ;
-      int x;
-     
-      printf ("\n%s (%d)  CC=%g n_cats=%d:\n", 
-	      var_get_name (vp->var), vp->base_subscript_long, vp->cc, vp->n_cats);
-
-      printf ("Reverse map\n");
-      for (x = 0 ; x < vp->n_cats; ++x)
-	{
-	  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 ("\nForward map\n");
-      for (node = hmap_first (m); node; node = hmap_next (m, node))
+      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)
 	{
-	  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);
+	  case_unref (iv->ccase);
 	}
+      hmap_destroy (&cat->iap[i].ivmap);
     }
 
-  assert (cat->n_vars <= cat->n_iap);
-
-  printf ("\n");
-  printf ("Number of categorical variables: %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);
+  /* 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;
+  struct interaction_value *iv = NULL;
   size_t hash = interaction_case_hash (iact, c);
 
-  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;
 }
 
 
 struct categoricals *
-categoricals_create (const struct interaction **inter, size_t n_inter,
+categoricals_create (struct interaction *const*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
@@ -246,13 +315,29 @@ categoricals_create (const struct interaction **inter, size_t n_inter,
   cat->aux1 = aux1;
   cat->aux2 = aux2;
 
-
   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;
+      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);
+	    }
+	}
     }
 
   return cat;
@@ -263,67 +348,94 @@ categoricals_create (const struct interaction **inter, size_t n_inter,
 void
 categoricals_update (struct categoricals *cat, const struct ccase *c)
 {
-  size_t i;
-  
+  int i;
+  struct variable_node *vn = NULL;
   const double weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
 
   assert (NULL == cat->reverse_variable_map_short);
   assert (NULL == cat->reverse_variable_map_long);
 
+  /* 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);
+	  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 ;
 
-      if ( interaction_case_is_missing (iact, c, cat->exclude))
-	continue;
+      //      if ( interaction_case_is_missing (iact, c, cat->exclude))
+      //         continue;
 
-      hash = interaction_case_hash (iact, c);
-      node = lookup_case (&cat->iap[i].map, iact, c);
+      size_t hash = interaction_case_hash (iact, c);
+      struct interaction_value *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;
-
-	  hmap_insert (&cat->iap[i].map, &node->node,  hash);
-	  cat->n_cats_total++;
-	  
-	  if ( 0 == cat->iap[i].n_cats)
-	    cat->n_vars++;
+	  node->cc = weight;
 
-	  node->subscript = cat->iap[i].n_cats++ ;
+	  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);
 	}
-
-      node->cc += weight;
+      else
+	{
+	  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);
+      	cat->update (node->user_data, cat->exclude, cat->wv, NULL, c, cat->aux1, cat->aux2);
     }
 }
 
-/* Return the number of categories (distinct values) for variable N */
+/* Return the number of categories (distinct values) for interction N */
 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);
+}
+
+
+size_t
+categoricals_df (const struct categoricals *cat, size_t n)
+{
+  return cat->iap[n].df;
 }
 
 
 /* Return the total number of categories */
 size_t
-categoricals_total (const struct categoricals *cat)
+categoricals_n_total (const struct categoricals *cat)
 {
+  assert (cat->reverse_variable_map_long);
+
   return cat->n_cats_total;
 }
 
+size_t
+categoricals_df_total (const struct categoricals *cat)
+{
+  return cat->df_sum;
+}
 
 /* This function must be called *before* any call to categoricals_get_*_by subscript and
  *after* all calls to categoricals_update */
@@ -338,47 +450,82 @@ categoricals_done (const struct categoricals *cat_)
   */
   struct categoricals *cat = CONST_CAST (struct categoricals *, cat_);
   int v;
+  int i;
   int idx_short = 0;
   int idx_long = 0;
+  cat->df_sum = 0;
+  cat->n_cats_total = 0;
+
+  /* Calculate the degrees of freedom, and the number of categories */
+  for (i = 0 ; i < cat->n_iap; ++i)
+    {
+      const struct interaction *iact = cat->iap[i].iact;
+
+      cat->iap[i].df = 1;
+      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].n_cats *= hmap_count (&vn->valmap);
+	}
+
+      cat->df_sum += cat->iap[i].df;
+      cat->n_cats_total += cat->iap[i].n_cats;
+    }
+
+
   cat->reverse_variable_map_short = pool_calloc (cat->pool,
-						 cat->n_cats_total - cat->n_vars,
+						 cat->df_sum,
 						 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)
+
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      int i;
-      struct interact_params *vp = &cat->iap[v];
-      int n_cats_total = categoricals_n_count (cat, v);
-      struct hmap_node *node ;
+      struct interaction_value *ivn = NULL;
+      int x = 0;
+      int ii;
+      struct interact_params *iap = &cat->iap[i];
 
-      vp->reverse_value_map = pool_calloc (cat->pool, n_cats_total, sizeof *vp->reverse_value_map);
+      iap->base_subscript_short = idx_short;
+      iap->base_subscript_long = idx_long;
 
-      vp->base_subscript_short = idx_short;
-      vp->base_subscript_long = idx_long;
+      iap->reverse_interaction_value_map = pool_calloc (cat->pool, iap->n_cats,
+							sizeof *iap->reverse_interaction_value_map);
 
-      for (node = hmap_first (&vp->map); node; node = hmap_next (&vp->map, node))
+      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->reverse_interaction_value_map[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->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 (i = 0; i < vp->n_cats - 1; ++i)
-	cat->reverse_variable_map_short[idx_short++] = v;
+      for (ii = 0; ii < iap->df; ++ii)
+	cat->reverse_variable_map_short[idx_short++] = i;
 
-      for (i = 0; i < vp->n_cats; ++i)
-	cat->reverse_variable_map_long[idx_long++] = v;
+      for (ii = 0; ii < iap->n_cats; ++ii)
+	cat->reverse_variable_map_long[idx_long++] = i;
     }
 
   assert (cat->n_vars <= cat->n_iap);
+
+  //  categoricals_dump (cat);
 }
 
 
@@ -387,7 +534,7 @@ reverse_variable_lookup_short (const struct categoricals *cat, int subscript)
 {
   assert (cat->reverse_variable_map_short);
   assert (subscript >= 0);
-  assert (subscript < cat->n_cats_total - cat->n_vars);
+  assert (subscript < cat->df_sum);
 
   return cat->reverse_variable_map_short[subscript];
 }
@@ -419,7 +566,10 @@ 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 value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_short];
+  const struct interaction_value *vn = vp->reverse_interaction_value_map [subscript - vp->base_subscript_short];
+
+  if ( vn == NULL)
+    return NULL;
 
   return vn->ccase;
 }
@@ -440,10 +590,13 @@ 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 value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_short];
-  return vn->cc;
-}
+  const struct interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_short];
+
+  if (iv == NULL)
+    return 0;
 
+  return iv->cc;
+}
 
 /* Returns unity if the value in case C at SUBSCRIPT is equal to the category
    for that subscript */
@@ -455,6 +608,9 @@ categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscr
 
   const struct ccase *c2 =  categoricals_get_case_by_subscript (cat, subscript);
 
+  if ( c2 == NULL)
+    return 0;
+
   return interaction_case_equal (iact, c, c2);
 }
 
@@ -462,28 +618,27 @@ categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscr
 size_t
 categoricals_get_n_variables (const struct categoricals *cat)
 {
+  printf ("%s\n", __FUNCTION__);
   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 value_node *vn = vp->reverse_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;
 }
 
-
 void *
 categoricals_get_user_data_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 value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_long];
-  return vn->user_data;
+  const struct interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
+  return iv->user_data;
 }