Categoricals: Use moments instead of keeping cc count ourselves

diff --git a/src/language/stats/oneway.c b/src/language/stats/oneway.c
index e88a9542ecd33855f4370c6b10cd76a8ef86ffc0..69570bbc6b9f2c76b396597387e7bdca5af3df2e 100644 (file)
--- a/src/language/stats/oneway.c
+++ b/src/language/stats/oneway.c
@@ -102,6 +102,15 @@ struct oneway_spec
 
 };
 
+/* Per category data */
+struct descriptive_data
+{
+  const struct variable *var;
+  struct moments1 *mom;
+
+  double minimum;
+  double maximum;
+};
 
 /* Workspace variable for each dependent variable */
 struct per_var_ws
@@ -114,7 +123,6 @@ struct per_var_ws
 
   int n_groups;
 
-  double cc;
   double mse;
 };
 
@@ -314,15 +322,6 @@ free_double (void *value_, const void *aux UNUSED)
 static void postcalc (const struct oneway_spec *cmd);
 static void  precalc (const struct oneway_spec *cmd);
 
-struct descriptive_data
-{
-  const struct variable *var;
-  struct moments1 *mom;
-
-  double minimum;
-  double maximum;
-};
-
 static struct descriptive_data *
 dd_create (const struct variable *var)
 {
@@ -348,7 +347,9 @@ makeit (void *aux1, void *aux2 UNUSED)
 }
 
 static void 
-updateit (void *user_data, const struct variable *wv, 
+updateit (void *user_data, 
+         enum mv_class exclude,
+         const struct variable *wv, 
          const struct variable *catvar UNUSED,
          const struct ccase *c,
          void *aux1, void *aux2)
@@ -359,6 +360,9 @@ updateit (void *user_data, const struct variable *wv,
 
   const union value *valx = case_data (c, varp);
 
+  if ( var_is_value_missing (varp, valx, exclude))
+    return;
+
   struct descriptive_data *dd_total = aux2;
 
   double weight = 1.0;
@@ -419,7 +423,6 @@ run_oneway (const struct oneway_spec *cmd,
       ws.vws[v].cov = covariance_2pass_create (1, &cmd->vars[v],
                                               cats, 
                                               cmd->wv, cmd->exclude);
-      ws.vws[v].cc = 0;
     }
 
   c = casereader_peek (input, 0);
@@ -469,7 +472,6 @@ run_oneway (const struct oneway_spec *cmd,
          {
            struct per_var_ws *pvw = &ws.vws[i];
 
-           pvw->cc += weight;
            covariance_accumulate_pass1 (pvw->cov, c);
          }
 
@@ -546,8 +548,13 @@ run_oneway (const struct oneway_spec *cmd,
       gsl_matrix *cm = covariance_calculate_unnormalized (pvw->cov);
       const struct categoricals *cats = covariance_get_categoricals (pvw->cov);
 
+      double n;
+      moments1_calculate (ws.dd_total[v]->mom, &n, NULL, NULL, NULL, NULL);
+
       pvw->sst = gsl_matrix_get (cm, 0, 0);
 
+      //      gsl_matrix_fprintf (stdout, cm, "%g ");
+
       reg_sweep (cm, 0);
 
       pvw->sse = gsl_matrix_get (cm, 0, 0);
@@ -556,7 +563,7 @@ run_oneway (const struct oneway_spec *cmd,
 
       pvw->n_groups = categoricals_total (cats);
 
-      pvw->mse = (pvw->sst - pvw->ssa) / (pvw->cc - pvw->n_groups);
+      pvw->mse = (pvw->sst - pvw->ssa) / (n - pvw->n_groups);
     }
 
   postcalc (cmd);
@@ -693,14 +700,12 @@ output_oneway (const struct oneway_spec *cmd, struct oneway_workspace *ws)
 
   show_anova_table (cmd, ws);
 
-
   if (ll_count (&cmd->contrast_list) > 0)
     {
       show_contrast_coeffs (cmd, ws);
       show_contrast_tests (cmd, ws);
     }
 
-
   /* Clean up */
   for (i = 0; i < cmd->n_vars; ++i )
     {
@@ -744,9 +749,12 @@ show_anova_table (const struct oneway_spec *cmd, const struct oneway_workspace *
 
   for (i = 0; i < cmd->n_vars; ++i)
     {
+      double n;
+      moments1_calculate (ws->dd_total[i]->mom, &n, NULL, NULL, NULL, NULL);
+
       const struct per_var_ws *pvw = &ws->vws[i];
       const double df1 = pvw->n_groups - 1;
-      const double df2 = pvw->cc - pvw->n_groups;
+      const double df2 = n - pvw->n_groups;
       const double msa = pvw->ssa / df1;
 
       const char *s = var_to_string (cmd->vars[i]);
@@ -769,7 +777,7 @@ show_anova_table (const struct oneway_spec *cmd, const struct oneway_workspace *
       /* Degrees of freedom */
       tab_fixed (t, 3, i * 3 + 1, 0, df1, 4, 0);
       tab_fixed (t, 3, i * 3 + 2, 0, df2, 4, 0);
-      tab_fixed (t, 3, i * 3 + 3, 0, pvw->cc - 1, 4, 0);
+      tab_fixed (t, 3, i * 3 + 3, 0, n - 1, 4, 0);
 
       /* Mean Squares */
       tab_double (t, 4, i * 3 + 1, TAB_RIGHT, msa, NULL);
@@ -985,6 +993,9 @@ show_homogeneity (const struct oneway_spec *cmd, const struct oneway_workspace *
 
   for (v = 0; v < cmd->n_vars; ++v)
     {
+      double n;
+      moments1_calculate (ws->dd_total[v]->mom, &n, NULL, NULL, NULL, NULL);
+
       const struct per_var_ws *pvw = &ws->vws[v];
       const struct categoricals *cats = covariance_get_categoricals (pvw->cov);
 
@@ -993,7 +1004,7 @@ show_homogeneity (const struct oneway_spec *cmd, const struct oneway_workspace *
       const char *s = var_to_string (var);
 
       const double df1 = pvw->n_groups - 1;
-      const double df2 = pvw->cc - pvw->n_groups;
+      const double df2 = n - pvw->n_groups;
       double F = gp->levene;
 
       tab_text (t, 0, v + 1, TAB_LEFT | TAT_TITLE, s);

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 1968b332f8257229d0d9cd25d4c601a3ed764426..576e19b1d322be30f2ee878aeda65e33083347d2 100644 (file)
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -88,16 +88,12 @@ struct categoricals
   /* 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;
   void *aux2;
@@ -119,7 +115,7 @@ categoricals_destroy ( struct categoricals *cat)
 }
 
 
-#if 0
+#if 1
 void
 categoricals_dump (const struct categoricals *cat)
 {
@@ -280,7 +276,7 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
       cat->vp[i].cc += weight;
 
       if (cat->update)
-       cat->update (node->user_data, cat->wv, var, c, cat->aux1, cat->aux2);
+       cat->update (node->user_data, cat->exclude, cat->wv, var, c, cat->aux1, cat->aux2);
     }
 }
 
@@ -341,6 +337,8 @@ categoricals_done (struct categoricals *cat)
     }
 
   assert (cat->n_vars <= cat->n_vp);
+
+  //  categoricals_dump (cat);
 }
 
 

diff --git a/src/math/categoricals.h b/src/math/categoricals.h
index 02fef13a303d17d717ef07df8b54c58a5302a2cc..09ced7d7acc2a9900fd67039697b98273aaafe0c 100644 (file)
--- a/src/math/categoricals.h
+++ b/src/math/categoricals.h
@@ -28,6 +28,7 @@ struct ccase;
 union value ;
 
 typedef void update_func (void *user_data,
+                         enum mv_class exclude,
                          const struct variable *wv, 
                          const struct variable *catvar,
                          const struct ccase *c,