/* 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
#include "gl/xalloc.h"
+#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 */
};
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
/* 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;
/* Missing values to be excluded */
enum mv_class exclude;
- /* Function to be called on each update */
- update_func *update;
-
- /* 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
/* 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);
}
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 exclude)
{
size_t i;
struct categoricals *cat = xmalloc (sizeof *cat);
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->payload = NULL;
+ cat->aux2 = NULL;
cat->iap = pool_calloc (cat->pool, cat->n_iap, sizeof *cat->iap);
{
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);
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);
{
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;
+
+ if ( interaction_case_is_missing (iact, c, cat->exclude))
+ continue;
- size_t hash = interaction_case_hash (iact, c, 0);
- struct interaction_value *node = lookup_case (&cat->iap[i].ivmap, iact, c);
+ hash = interaction_case_hash (iact, c, 0);
+ node = lookup_case (&cat->iap[i].ivmap, iact, c);
if ( NULL == node)
{
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
{
}
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)
+ cat->payload->update (cat->aux1, cat->aux2, node->user_data, c, cat->exclude, cat->wv);
+
}
}
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];
}
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;
}
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
/* 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 = xcalloc (iact->n_vars, sizeof (int));
+ 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;
+ cat->df_sum += cat->iap[i].df_prod [v - 1];
cat->n_cats_total += cat->iap[i].n_cats;
}
iap->reverse_interaction_value_map[ii] = NULL;
/* Populate the reverse variable maps. */
- for (ii = 0; ii < iap->df; ++ii)
+ 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)
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;
}
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)
{
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);
- 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;
- return interaction_case_equal (iact, c, c2);
+ 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];
+
+#if EFFECTS_CODING
+ if ( valn->index == df )
+ bin = -1.0;
+ else
+#endif
+ if ( valn->index != index )
+ bin = 0;
+
+ result *= bin;
+ }
+
+ return result;
}
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 interact_params *vp = &cat->iap[iact];
+ const struct interaction_value *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 = 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)
const struct interaction_value *iv = vp->reverse_interaction_value_map [subscript - vp->base_subscript_long];
return iv->user_data;
}
+
+
+\f
+
+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;
+}
projects / pspp-builds.git / blobdiff