Made GLM as well as ONEWAY work

[pspp] / src / language / stats / glm.c

diff --git a/src/language/stats/glm.c b/src/language/stats/glm.c
index 4ba93150c496db428ece611c79f7a149277e10b2..1eaf1a988a335264a5db70d9cd3bc137363c7602 100644 (file)
--- a/src/language/stats/glm.c
+++ b/src/language/stats/glm.c
@@ -54,16 +54,8 @@ struct glm_spec
   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;
 
@@ -104,18 +96,17 @@ static bool parse_design_spec (struct lexer *lexer, struct glm_spec *glm);
 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);
@@ -293,6 +284,8 @@ cmd_glm (struct lexer *lexer, struct dataset *ds)
 
   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);
 
@@ -303,6 +296,9 @@ error:
 
   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);
 
@@ -326,38 +322,33 @@ not_dropped (size_t j, size_t * dropped, size_t n_dropped)
 }
 
 static void
-get_ssq (struct covariance *cov, gsl_vector * ssq, const struct glm_spec *cmd)
+get_ssq (struct covariance *cov, gsl_vector *ssq, const struct glm_spec *cmd)
 {
-  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;
-
-  dropped = xcalloc (covariance_dim (cov), sizeof (*dropped));
-  vars = xcalloc (covariance_dim (cov), sizeof (*vars));
-  covariance_get_var_indices (cov, vars);
+  size_t *dropped = xcalloc (covariance_dim (cov), sizeof (*dropped));
+  const struct categoricals *cats = covariance_get_categoricals (cov);
 
-  for (k = 0; k < cmd->n_design_vars; k++)
+  for (k = 0; k < cmd->n_interactions; k++)
     {
-      n_dropped = 0;
-      for (i = 1; i < covariance_dim (cov); i++)
+      size_t n = 0;
+      size_t m = 0;
+      gsl_matrix *small_cov = NULL;
+      size_t n_dropped = 0;
+      for (i = cmd->n_dep_vars; i < covariance_dim (cov); i++)
        {
-         if (vars[i] == cmd->design_vars[k])
+         if (categoricals_get_interaction_by_subscript (cats, i - cmd->n_dep_vars)
+             == cmd->interactions[k])
            {
+             assert (n_dropped < covariance_dim (cov));
              dropped[n_dropped++] = i;
            }
        }
       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))
@@ -383,7 +374,6 @@ get_ssq (struct covariance *cov, gsl_vector * ssq, const struct glm_spec *cmd)
     }
 
   free (dropped);
-  free (vars);
   gsl_matrix_free (cm);
 }
 
@@ -437,8 +427,6 @@ run_glm (struct glm_spec *cmd, struct casereader *input,
     }
   casereader_destroy (reader);
 
-  categoricals_done (ws.cats);
-
   for (reader = input;
        (c = casereader_read (reader)) != NULL; case_unref (c))
     {
@@ -529,8 +517,7 @@ output_glm (const struct glm_spec *cmd, const struct glm_workspace *ws)
   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);
 
@@ -557,7 +544,7 @@ output_glm (const struct glm_spec *cmd, const struct glm_workspace *ws)
   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);
@@ -693,14 +680,10 @@ parse_design_interaction (struct lexer *lexer, struct glm_spec *glm, struct inte
 
   if ( lex_match (lexer, T_ASTERISK) || lex_match (lexer, T_BY))
     {
-      lex_error (lexer, "Interactions are not yet implemented"); return false;
+      // 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;
 }