#include <gsl/gsl_cdf.h>
#include <gsl/gsl_matrix.h>
+#include <gsl/gsl_combination.h>
#include <math.h>
#include "data/case.h"
size_t n_factor_vars;
const struct variable **factor_vars;
- /* In the current implementation, design_vars will
- normally be the same as factor_vars.
- This will change once interactions, nested variables
- and repeated measures become involved.
- */
- size_t n_design_vars;
- const struct variable **design_vars;
-
size_t n_interactions;
- const struct interaction **interactions;
+ struct interaction **interactions;
enum mv_class exclude;
gsl_vector *ssq;
};
+
+/* Default design: all possible interactions */
+static void
+design_full (struct glm_spec *glm)
+{
+ int sz;
+ int i = 0;
+ glm->n_interactions = (1 << glm->n_factor_vars) - 1;
+
+ glm->interactions = xcalloc (glm->n_interactions, sizeof *glm->interactions);
+
+ /* All subsets, with exception of the empty set, of [0, glm->n_factor_vars) */
+ for (sz = 1; sz <= glm->n_factor_vars; ++sz)
+ {
+ gsl_combination *c = gsl_combination_calloc (glm->n_factor_vars, sz);
+
+ do
+ {
+ struct interaction *iact = interaction_create (NULL);
+ int e;
+ for (e = 0 ; e < gsl_combination_k (c); ++e)
+ interaction_add_variable (iact, glm->factor_vars [gsl_combination_get (c, e)]);
+
+ glm->interactions[i++] = iact;
+ }
+ while (gsl_combination_next (c) == GSL_SUCCESS);
+
+ gsl_combination_free (c);
+ }
+}
+
static void output_glm (const struct glm_spec *,
const struct glm_workspace *ws);
static void run_glm (struct glm_spec *cmd, struct casereader *input,
int
cmd_glm (struct lexer *lexer, struct dataset *ds)
{
+ int i;
struct const_var_set *factors = NULL;
struct glm_spec glm;
bool design = false;
glm.dict = dataset_dict (ds);
glm.n_dep_vars = 0;
glm.n_factor_vars = 0;
- glm.n_design_vars = 0;
glm.n_interactions = 0;
glm.interactions = NULL;
glm.dep_vars = NULL;
glm.factor_vars = NULL;
- glm.design_vars = NULL;
glm.exclude = MV_ANY;
glm.intercept = true;
glm.wv = dict_get_weight (glm.dict);
if (! parse_design_spec (lexer, &glm))
goto error;
-
- if ( glm.n_interactions == 0)
- {
- msg (ME, _("One or more design variables must be given"));
- goto error;
- }
-
- design = true;
+
+ if (glm.n_interactions > 0)
+ design = true;
}
else
{
if ( ! design )
{
- lex_error (lexer, _("/DESIGN is mandatory in GLM"));
- goto error;
+ design_full (&glm);
}
{
const_var_set_destroy (factors);
free (glm.factor_vars);
+ for (i = 0 ; i < glm.n_interactions; ++i)
+ interaction_destroy (glm.interactions[i]);
free (glm.interactions);
free (glm.dep_vars);
const_var_set_destroy (factors);
free (glm.factor_vars);
+ for (i = 0 ; i < glm.n_interactions; ++i)
+ interaction_destroy (glm.interactions[i]);
+
free (glm.interactions);
free (glm.dep_vars);
const struct glm_spec *);
static bool
-not_dropped (size_t j, size_t * dropped, size_t n_dropped)
+not_dropped (size_t j, const size_t *dropped, size_t n_dropped)
{
size_t i;
return true;
}
-static void
-get_ssq (struct covariance *cov, gsl_vector * ssq, const struct glm_spec *cmd)
+/*
+ Do the variables in X->VARS constitute a proper
+ subset of the variables in Y->VARS?
+ */
+static bool
+is_subset (const struct interaction *x, const struct interaction *y)
{
- const struct variable **vars;
- gsl_matrix *small_cov = NULL;
- gsl_matrix *cm = covariance_calculate_unnormalized (cov);
size_t i;
size_t j;
- size_t k;
- size_t n;
- size_t m;
- size_t *dropped;
- size_t n_dropped;
+ size_t n = 0;
- dropped = xcalloc (covariance_dim (cov), sizeof (*dropped));
- vars = xcalloc (covariance_dim (cov), sizeof (*vars));
- covariance_get_var_indices (cov, vars);
-
- for (k = 0; k < cmd->n_design_vars; k++)
+ if (x->n_vars < y->n_vars)
{
- n_dropped = 0;
- for (i = 1; i < covariance_dim (cov); i++)
+ for (i = 0; i < x->n_vars; i++)
{
- if (vars[i] == cmd->design_vars[k])
+ for (j = 0; j < y->n_vars; j++)
{
- dropped[n_dropped++] = i;
+ if (x->vars [i] == y->vars [j])
+ {
+ n++;
+ }
}
}
- small_cov =
- gsl_matrix_alloc (cm->size1 - n_dropped, cm->size2 - n_dropped);
- gsl_matrix_set (small_cov, 0, 0, gsl_matrix_get (cm, 0, 0));
- n = 0;
- m = 0;
- for (i = 0; i < cm->size1; i++)
- {
- if (not_dropped (i, dropped, n_dropped))
+ }
+ if (n >= x->n_vars)
+ return true;
+ return false;
+}
+
+static bool
+drop_from_submodel (const struct interaction *x, const struct interaction *y)
+{
+ size_t i;
+ size_t j;
+ size_t n = 0;
+
+ if (is_subset (x, y))
+ return true;
+
+ for (i = 0; i < x->n_vars; i++)
+ for (j = 0; j < y->n_vars; j++)
+ {
+ if (x->vars [i] == y->vars [j])
+ n++;
+ }
+ if (n == x->n_vars)
+ {
+ return true;
+ }
+
+ return false;
+}
+
+static void
+fill_submatrix (gsl_matrix * cov, gsl_matrix * submatrix, size_t * dropped,
+ size_t n_dropped)
+{
+ size_t i;
+ size_t j;
+ size_t n = 0;
+ size_t m = 0;
+
+ for (i = 0; i < cov->size1; i++)
+ {
+ if (not_dropped (i, dropped, n_dropped))
+ {
+ m = 0;
+ for (j = 0; j < cov->size2; j++)
{
- m = 0;
- for (j = 0; j < cm->size2; j++)
+ if (not_dropped (j, dropped, n_dropped))
{
- if (not_dropped (j, dropped, n_dropped))
- {
- gsl_matrix_set (small_cov, n, m,
- gsl_matrix_get (cm, i, j));
- m++;
- }
- }
- n++;
+ gsl_matrix_set (submatrix, n, m,
+ gsl_matrix_get (cov, i, j));
+ m++;
+ }
}
+ n++;
}
- reg_sweep (small_cov, 0);
+ }
+}
+
+static void
+get_ssq (struct covariance *cov, gsl_vector *ssq, const struct glm_spec *cmd)
+{
+ gsl_matrix *cm = covariance_calculate_unnormalized (cov);
+ size_t i;
+ size_t k;
+ size_t *model_dropped = xcalloc (covariance_dim (cov), sizeof (*model_dropped));
+ size_t *submodel_dropped = xcalloc (covariance_dim (cov), sizeof (*submodel_dropped));
+ const struct categoricals *cats = covariance_get_categoricals (cov);
+
+ for (k = 0; k < cmd->n_interactions; k++)
+ {
+ gsl_matrix *model_cov = NULL;
+ gsl_matrix *submodel_cov = NULL;
+ size_t n_dropped_model = 0;
+ size_t n_dropped_submodel = 0;
+ for (i = cmd->n_dep_vars; i < covariance_dim (cov); i++)
+ {
+ const struct interaction * x =
+ categoricals_get_interaction_by_subscript (cats, i - cmd->n_dep_vars);
+ if (is_subset (cmd->interactions [k], x))
+ {
+ assert (n_dropped_model < covariance_dim (cov));
+ model_dropped[n_dropped_model++] = i;
+ }
+ if (drop_from_submodel (cmd->interactions [k], x))
+ {
+ assert (n_dropped_submodel < covariance_dim (cov));
+ submodel_dropped[n_dropped_submodel++] = i;
+ }
+ }
+ model_cov =
+ gsl_matrix_alloc (cm->size1 - n_dropped_model, cm->size2 - n_dropped_model);
+ gsl_matrix_set (model_cov, 0, 0, gsl_matrix_get (cm, 0, 0));
+ submodel_cov =
+ gsl_matrix_calloc (cm->size1 - n_dropped_submodel, cm->size2 - n_dropped_submodel);
+ fill_submatrix (cm, model_cov, model_dropped, n_dropped_model);
+ fill_submatrix (cm, submodel_cov, submodel_dropped, n_dropped_submodel);
+
+ reg_sweep (model_cov, 0);
+ reg_sweep (submodel_cov, 0);
gsl_vector_set (ssq, k + 1,
- gsl_matrix_get (small_cov, 0, 0)
- - gsl_vector_get (ssq, 0));
- gsl_matrix_free (small_cov);
+ gsl_matrix_get (submodel_cov, 0, 0)
+ - gsl_matrix_get (model_cov, 0, 0));
+ gsl_matrix_free (model_cov);
+ gsl_matrix_free (submodel_cov);
}
- free (dropped);
- free (vars);
+ free (model_dropped);
+ free (submodel_dropped);
gsl_matrix_free (cm);
}
struct glm_workspace ws;
struct covariance *cov;
- ws.cats = categoricals_create (cmd->design_vars, cmd->n_design_vars,
+
+ ws.cats = categoricals_create (cmd->interactions, cmd->n_interactions,
cmd->wv, cmd->exclude,
NULL, NULL, NULL, NULL);
}
casereader_destroy (reader);
- categoricals_done (ws.cats);
-
for (reader = input;
(c = casereader_read (reader)) != NULL; case_unref (c))
{
if (cmd->intercept)
df_corr += 1.0;
- for (f = 0; f < cmd->n_interactions; ++f)
- df_corr += categoricals_n_count (ws->cats, f) - 1.0;
+ df_corr += categoricals_df_total (ws->cats);
mse = gsl_vector_get (ws->ssq, 0) / (n_total - df_corr);
for (f = 0; f < cmd->n_interactions; ++f)
{
struct string str = DS_EMPTY_INITIALIZER;
- const double df = categoricals_n_count (ws->cats, f) - 1.0;
+ const double df = categoricals_df (ws->cats, f);
const double ssq = gsl_vector_get (ws->ssq, f + 1);
const double F = ssq / df / mse;
interaction_to_string (cmd->interactions[f], &str);
if ( lex_match (lexer, T_ASTERISK) || lex_match (lexer, T_BY))
{
- lex_error (lexer, "Interactions are not yet implemented"); return false;
return parse_design_interaction (lexer, glm, iact);
}
- glm->n_design_vars++;
- glm->design_vars = xrealloc (glm->design_vars, sizeof (*glm->design_vars) * glm->n_design_vars);
- glm->design_vars[glm->n_design_vars - 1] = v;
-
return true;
}
projects / pspp / blobdiff