ONEWAY: Fixed crash when the dependent variable contained only missing values.

[pspp-builds.git] / src / math / categoricals.c

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 4a7edecf673a565042940007d326f389c06f6bdf..d8606031c943b0925a1edc953e454693a9d8de8c 100644 (file)
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -32,6 +32,8 @@
 
 #include "gl/xalloc.h"
 
+#define EFFECTS_CODING 1
+
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
@@ -43,12 +45,11 @@ struct value_node
 
 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 */
 };
@@ -121,6 +122,8 @@ struct interact_params
      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;
 
@@ -261,6 +264,9 @@ categoricals_destroy (struct categoricals *cat)
        {
          case_unref (iv->ccase);
        }
+
+      free (cat->iap[i].enc_sum);
+      free (cat->iap[i].df_prod);
       hmap_destroy (&cat->iap[i].ivmap);
     }
 
@@ -434,7 +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);
+  if (!categoricals_is_complete (cat))
+    return 0;
 
   return cat->n_cats_total;
 }
@@ -445,9 +452,16 @@ 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
@@ -480,6 +494,9 @@ categoricals_done (const struct categoricals *cat_)
 
          struct variable_node *vn = lookup_variable (&cat->varmap, var, hash_pointer (var, 0));
 
+         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];
 
@@ -538,6 +555,30 @@ 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_code_for_case (cat, x, iv->ccase); \
+             iap->enc_sum [x - iap->base_subscript_short] += bin * iv->cc;
+           }
+       }
+    }
+
+  return true;
 }
 
 
@@ -571,21 +612,6 @@ 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)
 {
@@ -601,19 +627,13 @@ 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 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,
+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);
@@ -625,6 +645,9 @@ categoricals_get_binary_by_subscript (const struct categoricals *cat,
   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];
@@ -638,13 +661,14 @@ categoricals_get_binary_by_subscript (const struct categoricals *cat,
 
     double bin = 1.0;
 
+    const double df = iap->df_prod[v] / dfp;
+
     /* Translate the subscript into an index for the individual variable */
-    int index = (subscript - base_index) % cat->iap[i].df_prod[v];
-    if ( v > 0)
-      index /= cat->iap[i].df_prod[v - 1];
+    const int index = ((subscript - base_index) % iap->df_prod[v] ) / dfp;
+    dfp = iap->df_prod [v];
 
 #if EFFECTS_CODING
-    if ( valn->index == 0)
+    if ( valn->index == df )
       bin = -1.0;
     else 
 #endif
@@ -685,3 +709,7 @@ 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;
 }
+
+
+
+