/* A map indexed by a union values */
struct hmap map;
+ const struct variable *var;
+
int base_subscript;
/* The number of distinct values of this variable */
/* A map of values indexed by subscript */
const struct value_node **reverse_value_map;
+
+ /* Total of the weights of this variable */
+ double cc;
};
struct categoricals
{
- const struct variable **vars;
- size_t n_vars;
-
const struct variable *wv;
+ size_t n_vars;
+
/* An array of var_params */
struct var_params *vp;
{
const struct var_params *vp = &cat->vp[v];
const struct hmap *m = &vp->map;
- size_t width = var_get_width (cat->vars[v]);
+ // size_t width = var_get_width (vp->var);
struct hmap_node *node ;
int x;
- printf ("\n%s (%d):\n", var_get_name (cat->vars[v]), vp->base_subscript);
+ printf ("\n%s (%d) CC=%g:\n", var_get_name (vp->var), vp->base_subscript, vp->cc);
assert (vp->reverse_value_map);
struct string s;
const struct value_node *vn = vp->reverse_value_map[x];
ds_init_empty (&s);
- var_append_value_name (cat->vars[v], &vn->value, &s);
+ var_append_value_name (vp->var, &vn->value, &s);
printf ("Value for %d is %s\n", x, ds_cstr(&s));
ds_destroy (&s);
}
struct string s;
ds_init_empty (&s);
const struct value_node *vn = HMAP_DATA (node, struct value_node, node);
- var_append_value_name (cat->vars[v], &vn->value, &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);
size_t i;
struct categoricals *cat = xmalloc (sizeof *cat);
- cat->vars = v;
cat->n_vars = n_vars;
cat->wv = wv;
cat->n_cats_total = 0;
cat->vp = pool_calloc (cat->pool, n_vars, sizeof *cat->vp);
for (i = 0 ; i < cat->n_vars; ++i)
- hmap_init (&cat->vp[i].map);
+ {
+ hmap_init (&cat->vp[i].map);
+ cat->vp[i].var = v[i];
+ }
return cat;
}
for (i = 0 ; i < cat->n_vars; ++i)
{
- unsigned int width = var_get_width (cat->vars[i]);
- const union value *val = case_data (c, cat->vars[i]);
+ const struct variable *var = cat->vp[i].var;
+ unsigned int width = var_get_width (var);
+ const union value *val = case_data (c, var);
size_t hash = value_hash (val, width, 0);
- struct value_node *node = lookup_value (&cat->vp[i].map, cat->vars[i], val);
+ struct value_node *node = lookup_value (&cat->vp[i].map, var, val);
+
+
if ( NULL == node)
{
node = pool_malloc (cat->pool, sizeof *node);
}
node->cc += weight;
+ cat->vp[i].cc += weight;
}
}
int
categoricals_index (const struct categoricals *cat, size_t n, const union value *val)
{
- struct value_node *vn = lookup_value (&cat->vp[n].map, cat->vars[n], val);
+ struct value_node *vn = lookup_value (&cat->vp[n].map, cat->vp[n].var, val);
if ( vn == NULL)
return -1;
index = cat->reverse_variable_map[subscript];
- return cat->vars[index];
+ return cat->vp[index].var;
}