X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.c;h=e1bbe1f99b141994365d32de6f888a2dee2619e4;hb=f790dbda9d498eef9c9c0a49078adbeecf768d56;hp=80073deedbebfd2326f6d58c4eaca0650c7feb4b;hpb=bd009d07e62d87f89bd83415dd3ea194054d342e;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 80073deedb..e1bbe1f99b 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, 2012 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"
@@ -34,8 +35,6 @@
 
 #define CATEGORICALS_DEBUG 0
 
-#define EFFECTS_CODING 1
-
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
@@ -45,13 +44,15 @@ struct value_node
   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 */
+  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 */
+  double cc;                  /* Total of the weights of cases matching this interaction */
 
   void *user_data;            /* A pointer to data which the caller can store stuff */
 };
@@ -65,7 +66,7 @@ lookup_value (const struct hmap *map, const union value *val, unsigned int hash,
       if (value_equal (&vn->val, val, width))
 	break;
     }
-  
+
   return vn;
 }
 
@@ -78,18 +79,22 @@ struct variable_node
   int n_vals;                 /* Number of values for this variable */
 };
 
-static void UNUSED
-dump_interaction (const struct interaction *iact)
+
+/* Comparison function to sort value_nodes in ascending order */
+static int
+compare_value_node_3way (const void *vn1_, const void *vn2_, const void *aux)
 {
-  if (CATEGORICALS_DEBUG)
-    {
-      struct string str = DS_EMPTY_INITIALIZER;
-      interaction_to_string (iact, &str);
-      printf ("Interaction: %s\n", ds_cstr (&str));
-      ds_destroy (&str);
-    }
+  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)
 {
@@ -98,17 +103,17 @@ lookup_variable (const struct hmap *map, const struct variable *var, unsigned in
     {
       if (vn->var == var)
 	break;
-      
-      fprintf (stderr, "Warning: Hash table collision\n");
+
+      fprintf (stderr, "%s:%d Warning: Hash table collision\n", __FILE__, __LINE__);
     }
-  
+
   return vn;
 }
 
 
 struct interact_params
 {
-  /* A map indexed by a interaction_value */
+  /* A map of cases indexed by a interaction_value */
   struct hmap ivmap;
 
   const struct interaction *iact;
@@ -120,9 +125,9 @@ struct interact_params
   int n_cats;
 
   /* 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; 
+     These are the products of the degrees of freedom for the current
+     variable and all preceding variables */
+  int *df_prod;
 
   double *enc_sum;
 
@@ -177,16 +182,46 @@ 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;
+
+  /* Missing values in the factor variables to be excluded */
+  enum mv_class fctr_excl;
 
   const void *aux1;
   void *aux2;
 
+  bool sane;
+
   const struct payload *payload;
 };
 
-static void UNUSED
+
+bool
+categoricals_isbalanced (const struct categoricals *cat)
+{
+  int i;
+
+  for (i = 0 ; i < cat->n_iap; ++i)
+    {
+      int v;
+      const struct interact_params *iap = &cat->iap[i];
+
+      double oval = -1.0;
+      for (v = 0; v < hmap_count (&iap->ivmap); ++v)
+	{
+	  const struct interaction_value *iv = iap->reverse_interaction_value_map[v];
+	  if (oval == -1.0)
+	    oval = iv->cc;
+	  if (oval != iv->cc)
+	    return false;
+	}
+    }
+  return true;
+}
+
+
+static void
 categoricals_dump (const struct categoricals *cat)
 {
   if (CATEGORICALS_DEBUG)
@@ -207,8 +242,7 @@ categoricals_dump (const struct categoricals *cat)
 	}
       printf ("\n");
 
-
-      printf ("Number of interactions %d\n", cat->n_iap);
+      printf ("Number of interactions %zu\n", cat->n_iap);
       for (i = 0 ; i < cat->n_iap; ++i)
 	{
 	  int v;
@@ -219,24 +253,32 @@ categoricals_dump (const struct categoricals *cat)
 	  ds_init_empty (&str);
 	  interaction_to_string (iact, &str);
 
-	  printf ("\nInteraction: %s (n: %d); ", ds_cstr (&str), iap->n_cats);
+	  printf ("\nInteraction: \"%s\" (number of categories: %d); ", ds_cstr (&str), iap->n_cats);
 	  ds_destroy (&str);
-	  printf ("Base subscript: %d\n", iap->base_subscript_short);
+	  printf ("Base index (short/long): %d/%d\n", iap->base_subscript_short, iap->base_subscript_long);
 
 	  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)
 		{
 		  const struct variable *var = iact->vars[vv];
 		  const union value *val = case_data (iv->ccase, var);
-	      
-		  printf ("%g", val->f);
+		  unsigned int varhash = hash_pointer (var, 0);
+		  struct variable_node *vn = lookup_variable (&cat->varmap, var, varhash);
+
+		  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 (", ");
 		}
@@ -254,16 +296,15 @@ categoricals_destroy (struct categoricals *cat)
   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 0
-	  if (cat->payload)
-	    cat->payload->destroy (cat->aux1, iv->user_data);
-#endif
+	  if (cat->payload && cat->payload->destroy)
+	    cat->payload->destroy (cat->aux1, cat->aux2, iv->user_data);
 	  case_unref (iv->ccase);
 	}
 
@@ -304,14 +345,19 @@ lookup_case (const struct hmap *map, const struct interaction *iact, const struc
   return iv;
 }
 
+bool
+categoricals_sane (const struct categoricals *cat)
+{
+  return cat->sane;
+}
 
 struct categoricals *
 categoricals_create (struct interaction *const*inter, size_t n_inter,
-		     const struct variable *wv, enum mv_class exclude)
+		     const struct variable *wv, enum mv_class dep_excl, 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;
@@ -319,9 +365,11 @@ 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->dep_excl = dep_excl;
+  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);
 
@@ -359,7 +407,13 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 {
   int i;
   struct variable_node *vn = NULL;
-  const double weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
+  double weight;
+
+  if (NULL == cat)
+    return;
+
+  weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
+  weight = var_force_valid_weight (cat->wv, weight, NULL);
 
   assert (NULL == cat->reverse_variable_map_short);
   assert (NULL == cat->reverse_variable_map_long);
@@ -376,13 +430,14 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
       if (valn == NULL)
 	{
 	  valn = pool_malloc (cat->pool, sizeof *valn);
-	  valn->index = vn->n_vals++;
+	  valn->index = -1;
+	  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;
@@ -390,7 +445,7 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
       size_t hash;
       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, 0);
@@ -399,13 +454,12 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
       if ( NULL == node)
 	{
 	  node = pool_malloc (cat->pool, sizeof *node);
-
 	  node->ccase = case_ref (c);
 	  node->cc = weight;
 
 	  hmap_insert (&cat->iap[i].ivmap, &node->node, hash);
 
-	  if (cat->payload) 
+	  if (cat->payload)
 	    {
 	      node->user_data = cat->payload->create (cat->aux1, cat->aux2);
 	    }
@@ -418,10 +472,8 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 
       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);
 	}
-
     }
 }
 
@@ -454,6 +506,9 @@ categoricals_n_total (const struct categoricals *cat)
 size_t
 categoricals_df_total (const struct categoricals *cat)
 {
+  if (NULL == cat)
+    return 0;
+
   return cat->df_sum;
 }
 
@@ -466,7 +521,7 @@ categoricals_is_complete (const struct categoricals *cat)
 
 /* This function must be called *before* any call to categoricals_get_*_by subscript and
  *after* all calls to categoricals_update */
-bool
+void
 categoricals_done (const struct categoricals *cat_)
 {
   /* Implementation Note: Whilst this function is O(n) in cat->n_cats_total, in most
@@ -480,6 +535,10 @@ categoricals_done (const struct categoricals *cat_)
   int i;
   int idx_short = 0;
   int idx_long = 0;
+
+  if (NULL == cat)
+    return;
+
   cat->df_sum = 0;
   cat->n_cats_total = 0;
 
@@ -492,23 +551,51 @@ categoricals_done (const struct categoricals *cat_)
       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)
 	{
+	  int x;
 	  const struct variable *var = iact->vars[v];
 
 	  struct variable_node *vn = lookup_variable (&cat->varmap, var, hash_pointer (var, 0));
 
-	  if  (hmap_count (&vn->valmap) == 0)
-	    return false;
+	  struct value_node *valnd = NULL;
+	  struct value_node **array ;
+
+	  assert (vn->n_vals == hmap_count (&vn->valmap));
+
+	  if  (vn->n_vals == 0)
+	    {
+	      cat->sane = false;
+	      return;
+	    }
+
+	  /* Sort the VALMAP here */
+	  array = xcalloc (sizeof *array, vn->n_vals);
+	  x = 0;
+	  HMAP_FOR_EACH (valnd, struct value_node, node, &vn->valmap)
+	    {
+	      /* Note: This loop is probably superfluous, it could be done in the
+	       update stage (at the expense of a realloc) */
+	      array[x++] = valnd;
+	    }
+
+	  sort (array, vn->n_vals, sizeof (*array),
+		compare_value_node_3way, vn);
 
-	  cat->iap[i].df_prod[v] = df * (hmap_count (&vn->valmap) - 1);
+	  for (x = 0; x <  vn->n_vals; ++x)
+	    {
+	      struct value_node *vvv = array[x];
+	      vvv->index = x;
+	    }
+	  free (array);
+
+	  cat->iap[i].df_prod[v] = df * (vn->n_vals - 1);
       	  df = cat->iap[i].df_prod[v];
 
-	  cat->iap[i].n_cats *= hmap_count (&vn->valmap);
+	  cat->iap[i].n_cats *= vn->n_vals;
 	}
 
-      assert (v == iact->n_vars);
       if (v > 0)
 	cat->df_sum += cat->iap[i].df_prod [v - 1];
 
@@ -565,7 +652,7 @@ 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)
@@ -583,13 +670,15 @@ categoricals_done (const struct categoricals *cat_)
 	  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); \
+	      const double bin = categoricals_get_effects_code_for_case (cat, x, iv->ccase);
 	      iap->enc_sum [x - iap->base_subscript_short] += bin * iv->cc;
 	    }
+	  if (cat->payload && cat->payload->calculate)
+	    cat->payload->calculate (cat->aux1, cat->aux2, iv->user_data);
 	}
     }
 
-  return true;
+  cat->sane = true;
 }
 
 
@@ -641,11 +730,13 @@ categoricals_get_sum_by_subscript (const struct categoricals *cat, int subscript
   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
+static double
 categoricals_get_code_for_case (const struct categoricals *cat, int subscript,
-				const struct ccase *c)
+				const struct ccase *c,
+				bool effects_coding)
 {
   const struct interaction *iact = categoricals_get_interaction_by_subscript (cat, subscript);
 
@@ -678,14 +769,11 @@ categoricals_get_code_for_case (const struct categoricals *cat, int subscript,
       const int index = ((subscript - base_index) % iap->df_prod[v] ) / dfp;
       dfp = iap->df_prod [v];
 
-#if EFFECTS_CODING
-      if ( valn->index == df )
+      if (effects_coding && valn->index == df )
 	bin = -1.0;
-      else 
-#endif
-	if ( valn->index  != index )
-	  bin = 0;
-    
+      else if ( valn->index != index )
+	bin = 0;
+
       result *= bin;
     }
 
@@ -693,6 +781,28 @@ categoricals_get_code_for_case (const struct categoricals *cat, int subscript,
 }
 
 
+/* 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 categoricals_get_code_for_case (cat, subscript, c, true);
+}
+
+
 size_t
 categoricals_get_n_variables (const struct categoricals *cat)
 {
@@ -701,13 +811,19 @@ categoricals_get_n_variables (const struct categoricals *cat)
 }
 
 
-/* Return a case containing the set of values corresponding to 
+/* 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];
-  const struct interaction_value *vn = vp->reverse_interaction_value_map [n];
+
+  if ( n >= hmap_count (&vp->ivmap))
+    return NULL;
+
+  vn = vp->reverse_interaction_value_map [n];
 
   return vn->ccase;
 }
@@ -717,7 +833,12 @@ 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 = vp->reverse_interaction_value_map [n];
+  const struct interaction_value *iv ;
+
+  if ( n >= hmap_count (&vp->ivmap))
+    return NULL;
+
+  iv = vp->reverse_interaction_value_map [n];
 
   return iv->user_data;
 }