X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.c;h=586037b585c3a7d91aabf94da12e61a35d0ad8dc;hb=59c87092f400493fbc95ee72e5cd8bf02468ce01;hp=81c673b3fc590e02cdd522e192b1b85326a06ff8;hpb=dc29c57e7908b0bdc6ab84f8aa4b7cb8a780026c;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 81c673b3fc..586037b585 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 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
@@ -32,22 +32,26 @@
 
 #include "gl/xalloc.h"
 
+#define CATEGORICALS_DEBUG 0
+
+#define EFFECTS_CODING 1
+
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
 
   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 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 */
 
-  /* Total of the weights of cases matching this interaction */
-  double cc; 
+  double cc;        /* Total of the weights of cases matching this interaction */
 
   void *user_data;            /* A pointer to data which the caller can store stuff */
 };
@@ -71,20 +75,9 @@ 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)
-{
-  struct string str = DS_EMPTY_INITIALIZER;
-  interaction_to_string (iact, &str);
-  printf ("Interaction: %s\n", ds_cstr (&str));
-  ds_destroy (&str);
-}
-#endif
-
-
 static struct variable_node *
 lookup_variable (const struct hmap *map, const struct variable *var, unsigned int hash)
 {
@@ -114,8 +107,12 @@ 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;
@@ -158,7 +155,7 @@ struct categoricals
 
   /* 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;
 
   /* Like the above, but uses all N variables */
@@ -168,79 +165,78 @@ struct categoricals
 
   struct pool *pool;
 
-  /* Missing values to be excluded */
-  enum mv_class exclude;
+  /* Missing values in the dependent varirable to be excluded */
+  enum mv_class dep_excl;
 
-  /* Function to be called on each update */
-  update_func *update;
+  /* Missing values in the factor variables to be excluded */
+  enum mv_class fctr_excl;
 
-  /* Function specified by the caller to create user_data */
-  user_data_create_func *user_data_create;
-
-  /* Auxilliary data to be passed to update and user_data_create_func*/
-  void *aux1;
+  const void *aux1;
   void *aux2;
+
+  const struct payload *payload;
 };
 
-#if 0
 static void
 categoricals_dump (const struct categoricals *cat)
 {
-  int i;
-
-  printf ("Reverse Variable Map (short):\n");
-  for (i = 0; i < cat->df_sum; ++i)
+  if (CATEGORICALS_DEBUG)
     {
-      printf (" %d", cat->reverse_variable_map_short[i]);
-    }
-  printf ("\n");
+      int i;
 
-  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 ("Reverse Variable Map (short):\n");
+      for (i = 0; i < cat->df_sum; ++i)
+	{
+	  printf (" %d", cat->reverse_variable_map_short[i]);
+	}
+      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)
-    {
-      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 ("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;
 
-      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);
+	  ds_init_empty (&str);
+	  interaction_to_string (iact, &str);
 
-      printf ("\t(");
-      for (v = 0; v < hmap_count (&iap->ivmap); ++v)
-	{
-	  int vv;
-	  const struct interaction_value *iv = iap->reverse_interaction_value_map[v];
-	  
-	  if (v > 0)  printf ("   ");
-	  printf ("{");
-	  for (vv = 0; vv < iact->n_vars; ++vv)
+	  printf ("\nInteraction: %s (n: %d); ", ds_cstr (&str), iap->n_cats);
+	  ds_destroy (&str);
+	  printf ("Base subscript: %d\n", iap->base_subscript_short);
+
+	  printf ("\t(");
+	  for (v = 0; v < hmap_count (&iap->ivmap); ++v)
 	    {
-	      const struct variable *var = iact->vars[vv];
-	      const union value *val = case_data (iv->ccase, var);
+	      int vv;
+	      const struct interaction_value *iv = iap->reverse_interaction_value_map[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);
 	      
-	      printf ("%g", val->f);
-	      if (vv < iact->n_vars - 1)
-		printf (", ");
+		  printf ("%g", val->f);
+		  if (vv < iact->n_vars - 1)
+		    printf (", ");
+		}
+	      printf ("}");
 	    }
-	  printf ("}");
+	  printf (")\n");
 	}
-      printf (")\n");
     }
 }
-#endif
 
 void
 categoricals_destroy (struct categoricals *cat)
@@ -255,9 +251,17 @@ categoricals_destroy (struct categoricals *cat)
       /* 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 0
+	  if (cat->payload)
+	    cat->payload->destroy (cat->aux1, iv->user_data);
+#endif
 	  case_unref (iv->ccase);
 	}
+
+      free (cat->iap[i].enc_sum);
+      free (cat->iap[i].df_prod);
       hmap_destroy (&cat->iap[i].ivmap);
+      interaction_destroy (cat->iap[i].iact);
     }
 
   /* Interate over each variable and delete its value map */
@@ -295,10 +299,7 @@ lookup_case (const struct hmap *map, const struct interaction *iact, const struc
 
 struct categoricals *
 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
-		     )
+		     const struct variable *wv, enum mv_class dep_excl, enum mv_class fctr_excl)
 {
   size_t i;
   struct categoricals *cat = xmalloc (sizeof *cat);
@@ -310,12 +311,10 @@ categoricals_create (struct interaction *const*inter, size_t n_inter,
   cat->reverse_variable_map_short = NULL;
   cat->reverse_variable_map_long = NULL;
   cat->pool = pool_create ();
-  cat->exclude = exclude;
-  cat->update = update;
-  cat->user_data_create = udf;
-
-  cat->aux1 = aux1;
-  cat->aux2 = aux2;
+  cat->dep_excl = dep_excl;
+  cat->fctr_excl = fctr_excl;
+  cat->payload = NULL;
+  cat->aux2 = NULL;
 
   cat->iap = pool_calloc (cat->pool, cat->n_iap, sizeof *cat->iap);
 
@@ -335,6 +334,7 @@ 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);
@@ -369,6 +369,7 @@ 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);
@@ -379,11 +380,14 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
     {
       const struct interaction *iact = cat->iap[i].iact;
 
-      //      if ( interaction_case_is_missing (iact, c, cat->exclude))
-      //         continue;
+      size_t hash;
+      struct interaction_value *node;
 
-      size_t hash = interaction_case_hash (iact, c, 0);
-      struct interaction_value *node = lookup_case (&cat->iap[i].ivmap, iact, c);
+      if ( interaction_case_is_missing (iact, c, cat->fctr_excl))
+	continue;
+
+      hash = interaction_case_hash (iact, c, 0);
+      node = lookup_case (&cat->iap[i].ivmap, iact, c);
 
       if ( NULL == node)
 	{
@@ -394,8 +398,10 @@ 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->payload) 
+	    {
+	      node->user_data = cat->payload->create (cat->aux1, cat->aux2);
+	    }
 	}
       else
 	{
@@ -403,8 +409,12 @@ 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->payload)
+	{
+	  double weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
+	  cat->payload->update (cat->aux1, cat->aux2, node->user_data, c, weight);
+	}
+
     }
 }
 
@@ -419,7 +429,8 @@ 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];
 }
 
 
@@ -427,7 +438,8 @@ categoricals_df (const struct categoricals *cat, size_t n)
 size_t
 categoricals_n_total (const struct categoricals *cat)
 {
-  assert (cat->reverse_variable_map_long);
+  if (!categoricals_is_complete (cat))
+    return 0;
 
   return cat->n_cats_total;
 }
@@ -438,9 +450,16 @@ categoricals_df_total (const struct categoricals *cat)
   return cat->df_sum;
 }
 
+bool
+categoricals_is_complete (const struct categoricals *cat)
+{
+  return (NULL != cat->reverse_variable_map_short);
+}
+
+
 /* This function must be called *before* any call to categoricals_get_*_by subscript and
  *after* all calls to categoricals_update */
-void
+bool
 categoricals_done (const struct categoricals *cat_)
 {
   /* Implementation Note: Whilst this function is O(n) in cat->n_cats_total, in most
@@ -460,22 +479,32 @@ 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 = iact->n_vars ? xcalloc (iact->n_vars, sizeof (int)) : NULL;
 
+      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;
+	  if  (hmap_count (&vn->valmap) == 0)
+	    return false;
+
+	  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;
+      assert (v == iact->n_vars);
+      if (v > 0)
+	cat->df_sum += cat->iap[i].df_prod [v - 1];
+
       cat->n_cats_total += cat->iap[i].n_cats;
     }
 
@@ -504,6 +533,7 @@ categoricals_done (const struct categoricals *cat_)
       HMAP_FOR_EACH (ivn, struct interaction_value, node, &iap->ivmap)
 	{
 	  iap->reverse_interaction_value_map[x++] = ivn;
+
 	}
 
       assert (x <= iap->n_cats);
@@ -517,8 +547,11 @@ categoricals_done (const struct categoricals *cat_)
 	iap->reverse_interaction_value_map[ii] = NULL;
 
       /* Populate the reverse variable maps. */
-      for (ii = 0; ii < iap->df; ++ii)
-	cat->reverse_variable_map_short[idx_short++] = i;
+      if (iap->df_prod)
+	{
+	  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)
 	cat->reverse_variable_map_long[idx_long++] = i;
@@ -526,7 +559,33 @@ categoricals_done (const struct categoricals *cat_)
 
   assert (cat->n_vars <= cat->n_iap);
 
-  //  categoricals_dump (cat);
+  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->reverse_interaction_value_map[y];
+	  for (x = iap->base_subscript_short; x < iap->base_subscript_short + df ;++x)
+	    {
+	      const double bin = categoricals_get_code_for_case (cat, x, iv->ccase); \
+	      iap->enc_sum [x - iap->base_subscript_short] += bin * iv->cc;
+	    }
+	  if (cat->payload && cat->payload->destroy)
+	    cat->payload->destroy (cat->aux1, cat->aux2, iv->user_data);
+	}
+    }
+
+  return true;
 }
 
 
@@ -560,22 +619,6 @@ 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)
-{
-  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;
-}
-
-
 double
 categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscript)
 {
@@ -591,28 +634,58 @@ 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];
-
-  if (iv == NULL)
-    return 0;
-
-  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,
-				      const struct ccase *c)
+categoricals_get_code_for_case (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);
 
-  if ( c2 == NULL)
-    return 0;
+  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;
+
+      /* 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];
 
-  return interaction_case_equal (iact, c, c2);
+#if EFFECTS_CODING
+      if ( valn->index == df )
+	bin = -1.0;
+      else 
+#endif
+	if ( valn->index  != index )
+	  bin = 0;
+    
+      result *= bin;
+    }
+
+  return result;
 }
 
 
@@ -623,6 +696,41 @@ categoricals_get_n_variables (const struct categoricals *cat)
   return cat->n_vars;
 }
 
+
+/* 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->reverse_interaction_value_map [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->reverse_interaction_value_map [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)
@@ -643,3 +751,15 @@ categoricals_get_user_data_by_category (const struct categoricals *cat, int subs
   const struct interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
   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;
+}