X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Foneway.q;h=ff6b79b66e9d322eac109ce062ac5806cf8bfef5;hb=d2f8593a1f1d39a3264682af0da898a3d67b68cf;hp=e26402b18f081eb5cfe28474b74cf78cc3bfa059;hpb=f98e66bead58bd287261d7bc14dab07515498f55;p=pspp

diff --git a/src/oneway.q b/src/oneway.q
index e26402b18f..ff6b79b66e 100644
--- a/src/oneway.q
+++ b/src/oneway.q
@@ -39,7 +39,7 @@
 #include "vfm.h"
 #include "hash.h"
 #include "casefile.h"
-#include "oneway.h"
+#include "group_proc.h"
 #include "group.h"
 #include "levene.h"
 
@@ -84,16 +84,23 @@ static int ostensible_number_of_groups=-1;
 static is_missing_func value_is_missing;
 
 
-static void calculate(const struct casefile *cf, void *_mode);
+static void run_oneway(const struct casefile *cf, void *_mode);
 
 
 /* Routines to show the output tables */
 static void show_anova_table(void);
 static void show_descriptives(void);
 static void show_homogeneity(void);
-static void show_contrast_coeffs(void);
-static void show_contrast_tests(void);
 
+static void show_contrast_coeffs(short *);
+static void show_contrast_tests(short *);
+
+
+enum stat_table_t {STAT_DESC = 1, STAT_HOMO = 2};
+
+static enum stat_table_t stat_tables ;
+
+void output_oneway(void);
 
 
 int
@@ -110,7 +117,40 @@ cmd_oneway(void)
   else
     value_is_missing = is_missing;
 
-  multipass_procedure_with_splits (calculate, &cmd);
+  /* What statistics were requested */
+  if ( cmd.sbc_statistics ) 
+    {
+
+      for (i = 0 ; i < ONEWAY_ST_count ; ++i ) 
+	{
+	  if  ( ! cmd.a_statistics[i]  ) continue;
+
+	  switch (i) {
+	  case ONEWAY_ST_DESCRIPTIVES:
+	    stat_tables |= STAT_DESC;
+	    break;
+	  case ONEWAY_ST_HOMOGENEITY:
+	    stat_tables |= STAT_HOMO;
+	    break;
+	  }
+	}
+    }
+
+  multipass_procedure_with_splits (run_oneway, &cmd);
+
+
+  return CMD_SUCCESS;
+}
+
+
+void
+output_oneway(void)
+{
+
+  int i;
+  short *bad_contrast ; 
+
+  bad_contrast = xmalloc ( sizeof (short) * cmd.sbc_contrast );
 
   /* Check the sanity of the given contrast values */
   for (i = 0 ; i < cmd.sbc_contrast ; ++i ) 
@@ -118,12 +158,14 @@ cmd_oneway(void)
       int j;
       double sum = 0;
 
+      bad_contrast[i] = 0;
       if ( subc_list_double_count(&cmd.dl_contrast[i]) != 
 	   ostensible_number_of_groups )
 	{
-	  msg(SE, 
+	  msg(SW, 
 	      _("Number of contrast coefficients must equal the number of groups"));
-	  return CMD_FAILURE;
+	  bad_contrast[i] = 1;
+	  continue;
 	}
 
       for (j=0; j < ostensible_number_of_groups ; ++j )
@@ -133,53 +175,38 @@ cmd_oneway(void)
 	msg(SW,_("Coefficients for contrast %d do not total zero"),i + 1);
     }
 
+  if ( stat_tables & STAT_DESC ) 
+    show_descriptives();
 
-
-  /* Show the statistics tables */
-  if ( cmd.sbc_statistics ) 
-    {
-    for (i = 0 ; i < ONEWAY_ST_count ; ++i ) 
-      {
-	if  ( ! cmd.a_statistics[i]  ) continue;
-
-	switch (i) {
-	case ONEWAY_ST_DESCRIPTIVES:
-	  show_descriptives();
-	  break;
-	case ONEWAY_ST_HOMOGENEITY:
-	  show_homogeneity();
-	  break;
-	}
-      }
-  }
-
+  if ( stat_tables & STAT_HOMO )
+    show_homogeneity();
 
   show_anova_table();
      
-  if (cmd.sbc_contrast)
+  if (cmd.sbc_contrast )
     {
-      show_contrast_coeffs();
-      show_contrast_tests();
+      show_contrast_coeffs(bad_contrast);
+      show_contrast_tests(bad_contrast);
     }
 
 
+  free(bad_contrast);
+
   /* Clean up */
   for (i = 0 ; i < n_vars ; ++i ) 
     {
-      struct hsh_table *group_hash = vars[i]->p.ww.group_hash;
+      struct hsh_table *group_hash = vars[i]->p.grp_data.group_hash;
 
       hsh_destroy(group_hash);
     }
 
   hsh_destroy(global_group_hash);
 
-  return CMD_SUCCESS;
 }
 
 
 
 
-
 /* Parser for the variables sub command */
 static int
 oneway_custom_variables(struct cmd_oneway *cmd UNUSED)
@@ -254,8 +281,8 @@ show_anova_table(void)
 
   for ( i=0 ; i < n_vars ; ++i ) 
     {
-      struct group_statistics *totals = &vars[i]->p.ww.ugs;
-      struct hsh_table *group_hash = vars[i]->p.ww.group_hash;
+      struct group_statistics *totals = &vars[i]->p.grp_data.ugs;
+      struct hsh_table *group_hash = vars[i]->p.grp_data.group_hash;
       struct hsh_iterator g;
       struct group_statistics *gs;
       double ssa=0;
@@ -283,11 +310,11 @@ show_anova_table(void)
 
       {
 	const double sst = totals->ssq - ( totals->sum * totals->sum) / totals->n ;
-	const double df1 = vars[i]->p.ww.n_groups - 1;
-	const double df2 = totals->n - vars[i]->p.ww.n_groups ;
+	const double df1 = vars[i]->p.grp_data.n_groups - 1;
+	const double df2 = totals->n - vars[i]->p.grp_data.n_groups ;
 	const double msa = ssa / df1;
 	
-	vars[i]->p.ww.mse  = (sst - ssa) / df2;
+	vars[i]->p.grp_data.mse  = (sst - ssa) / df2;
 	
 	
 	/* Sums of Squares */
@@ -303,11 +330,11 @@ show_anova_table(void)
 
 	/* Mean Squares */
 	tab_float (t, 4, i * 3 + 1, TAB_RIGHT, msa, 8, 3);
-	tab_float (t, 4, i * 3 + 2, TAB_RIGHT, vars[i]->p.ww.mse, 8, 3);
+	tab_float (t, 4, i * 3 + 2, TAB_RIGHT, vars[i]->p.grp_data.mse, 8, 3);
 	
 
 	{ 
-	  const double F = msa/vars[i]->p.ww.mse ;
+	  const double F = msa/vars[i]->p.grp_data.mse ;
 
 	  /* The F value */
 	  tab_float (t, 5, i * 3 + 1, 0,  F, 8, 3);
@@ -345,7 +372,7 @@ show_descriptives(void)
 
 
   for ( v = 0 ; v < n_vars ; ++v ) 
-    n_rows += vars[v]->p.ww.n_groups + 1;
+    n_rows += vars[v]->p.grp_data.n_groups + 1;
 
   t = tab_create (n_cols,n_rows,0);
   tab_headers (t, 2, 0, 2, 0);
@@ -392,12 +419,12 @@ show_descriptives(void)
 
       struct hsh_iterator g;
       struct group_statistics *gs;
-      struct group_statistics *totals = &vars[v]->p.ww.ugs; 
+      struct group_statistics *totals = &vars[v]->p.grp_data.ugs; 
 
       int count = 0 ;      
       char *s = (vars[v]->label) ? vars[v]->label : vars[v]->name;
 
-      struct hsh_table *group_hash = vars[v]->p.ww.group_hash;
+      struct hsh_table *group_hash = vars[v]->p.grp_data.group_hash;
 
 
       tab_text (t, 0, row, TAB_LEFT | TAT_TITLE, s);
@@ -480,7 +507,7 @@ show_descriptives(void)
       tab_float(t, 8, row + count, 0,  totals->minimum, 8, 2); 
       tab_float(t, 9, row + count, 0,  totals->maximum, 8, 2); 
 
-      row += vars[v]->p.ww.n_groups + 1;
+      row += vars[v]->p.grp_data.n_groups + 1;
     }
 
 
@@ -526,9 +553,23 @@ show_homogeneity(void)
 
   for ( v=0 ; v < n_vars ; ++v ) 
     {
-      char *s = (vars[v]->label) ? vars[v]->label : vars[v]->name;
+      double F;
+      const struct variable *var = vars[v];
+      const char *s = (var->label) ? var->label : var->name;
+      const struct group_statistics *totals = &var->p.grp_data.ugs;
+
+      const double df1 = var->p.grp_data.n_groups - 1;
+      const double df2 = totals->n - var->p.grp_data.n_groups ;
 
       tab_text (t, 0, v + 1, TAB_LEFT | TAT_TITLE, s);
+
+      F = var->p.grp_data.levene;
+      tab_float (t, 1, v + 1, TAB_RIGHT, F, 8,3);
+      tab_float (t, 2, v + 1, TAB_RIGHT, df1 ,8,0);
+      tab_float (t, 3, v + 1, TAB_RIGHT, df2 ,8,0);
+
+      /* Now the significance */
+      tab_float (t, 4, v + 1, TAB_RIGHT,gsl_cdf_fdist_Q(F,df1,df2), 8, 3);
     }
 
   tab_submit (t);
@@ -539,7 +580,7 @@ show_homogeneity(void)
 
 /* Show the contrast coefficients table */
 static void 
-show_contrast_coeffs(void)
+show_contrast_coeffs(short *bad_contrast)
 {
   char *s;
   int n_cols = 2 + ostensible_number_of_groups;
@@ -599,6 +640,7 @@ show_contrast_coeffs(void)
       int i;
       char *lab;
 
+
       lab = val_labs_find(indep_var->val_labs,*group_value);
   
       if ( lab ) 
@@ -610,10 +652,15 @@ show_contrast_coeffs(void)
 
       for (i = 0 ; i < cmd.sbc_contrast ; ++i ) 
 	{
+
 	  tab_text(t, 1, i + 2, TAB_CENTER | TAT_PRINTF, "%d", i + 1);
-	  tab_text(t, count + 2, i + 2, TAB_RIGHT | TAT_PRINTF, "%g", 
-		   subc_list_double_at(&cmd.dl_contrast[i],count)
-		   );
+
+	  if ( bad_contrast[i] ) 
+	    tab_text(t, count + 2, i + 2, TAB_RIGHT, "?" );
+	  else
+	    tab_text(t, count + 2, i + 2, TAB_RIGHT | TAT_PRINTF, "%g", 
+		     subc_list_double_at(&cmd.dl_contrast[i],count)
+		     );
 	}
 	  
       count++ ; 
@@ -626,7 +673,7 @@ show_contrast_coeffs(void)
 
 /* Show the results of the contrast tests */
 static void 
-show_contrast_tests(void)
+show_contrast_tests(short *bad_contrast)
 {
   int v;
   int n_cols = 8;
@@ -680,14 +727,30 @@ show_contrast_tests(void)
 	  int ci;
 	  double contrast_value = 0.0;
 	  double coef_msq = 0.0;
-	  struct oneway_proc *ww = &vars[v]->p.ww ;
-	  struct hsh_table *group_hash = ww->group_hash;
+	  struct group_proc *grp_data = &vars[v]->p.grp_data ;
+	  struct hsh_table *group_hash = grp_data->group_hash;
 	  struct hsh_iterator g;
 	  struct group_statistics *gs;
 
 	  double T;
 	  double std_error_contrast ;
 	  double df;
+	  double sec_vneq=0.0;
+
+
+	  /* Note: The calculation of the degrees of freedom in the variances 
+	     not  equal case is painfull!!
+	     The following formula may help to understand it:
+	     \frac{\left(\sum_{i=1}^k{c_i^2\frac{s_i^2}{n_i}}\right)^2}
+	     {
+	     \sum_{i=1}^k\left(
+                \frac{\left(c_i^2\frac{s_i^2}{n_i}\right)^2}  {n_i-1}
+	     \right)
+	     }
+	  */
+
+	  double df_denominator = 0.0;
+	  double df_numerator = 0.0;
 
 	  
 	  if ( i == 0 ) 
@@ -708,6 +771,10 @@ show_contrast_tests(void)
 	  tab_text (t,  2, (v * lines_per_variable) + i + 1 + cmd.sbc_contrast,
 		    TAB_CENTER | TAT_TITLE | TAT_PRINTF, "%d",i+1);
 
+
+	  if ( bad_contrast[i]) 
+	    continue;
+
 	  /* FIXME: Potential danger here.
 	     We're ASSUMING THE array is in the order corresponding to the 
 	     hash order. */
@@ -715,21 +782,32 @@ show_contrast_tests(void)
 	       gs != 0;
 	       ++ci, gs = hsh_next(group_hash,&g))
 	    {
+
 	      const double coef = subc_list_double_at(&cmd.dl_contrast[i],ci);
+	      const double winv = (gs->std_dev * gs->std_dev) / gs->n;
 
 	      contrast_value += coef * gs->mean;
 
 	      coef_msq += (coef * coef) / gs->n ; 
+
+	      sec_vneq += (coef * coef) * (gs->std_dev * gs->std_dev ) /gs->n ;
+
+	      df_numerator += (coef * coef) * winv;
+	      df_denominator += pow2((coef * coef) * winv) / (gs->n - 1);
+
 	    }
+	  sec_vneq = sqrt(sec_vneq);
+
+	  df_numerator = pow2(df_numerator);
 
 	  tab_float (t,  3, (v * lines_per_variable) + i + 1, 
 		     TAB_RIGHT, contrast_value, 8,2);
 
-	  tab_float (t,  3, (v * lines_per_variable) + i + 1 + cmd.sbc_contrast,
+	  tab_float (t,  3, (v * lines_per_variable) + i + 1 + 
+		     cmd.sbc_contrast,
 		     TAB_RIGHT, contrast_value, 8,2);
 
-
-	  std_error_contrast = sqrt(vars[v]->p.ww.mse * coef_msq);
+	  std_error_contrast = sqrt(vars[v]->p.grp_data.mse * coef_msq);
 
 	  /* Std. Error */
 	  tab_float (t,  4, (v * lines_per_variable) + i + 1, 
@@ -744,7 +822,7 @@ show_contrast_tests(void)
 		     TAB_RIGHT, T,
 		     8,3);
 
-	  df = ww->ugs.n - ww->n_groups;
+	  df = grp_data->ugs.n - grp_data->n_groups;
 
 	  /* Degrees of Freedom */
 	  tab_float (t,  6, (v * lines_per_variable) + i + 1, 
@@ -757,51 +835,54 @@ show_contrast_tests(void)
 		     TAB_RIGHT,  2 * gsl_cdf_tdist_Q(T,df),
 		     8,3);
 
-	}
 
-      if ( v > 0 ) 
-	tab_hline(t, TAL_1, 0, n_cols - 1, (v * lines_per_variable) + 1);
-    }
+	  /* Now for the Variances NOT Equal case */
 
-  tab_submit (t);
+	  /* Std. Error */
+	  tab_float (t,  4, 
+		     (v * lines_per_variable) + i + 1 + cmd.sbc_contrast, 
+		     TAB_RIGHT, sec_vneq,
+		     8,3);
 
-}
 
+	  T = contrast_value / sec_vneq;
+	  tab_float (t,  5, 
+		     (v * lines_per_variable) + i + 1 + cmd.sbc_contrast, 
+		     TAB_RIGHT, T,
+		     8,3);
 
-/* ONEWAY ANOVA Calculations */
 
-static void  postcalc (  struct cmd_oneway *cmd UNUSED );
+	  df = df_numerator / df_denominator;
 
-static void  precalc ( struct cmd_oneway *cmd UNUSED );
+	  tab_float (t,  6, 
+		     (v * lines_per_variable) + i + 1 + cmd.sbc_contrast, 
+		     TAB_RIGHT, df,
+		     8,3);
 
-int  compare_group_id (const struct group_statistics *a, 
-		       const struct group_statistics *b, int width);
+	  /* The Significance */
 
-unsigned int hash_group_id(const struct group_statistics *v, int width);
+	  tab_float (t, 7, (v * lines_per_variable) + i + 1 + cmd.sbc_contrast,
+		     TAB_RIGHT,  2 * gsl_cdf_tdist_Q(T,df),
+		     8,3);
 
-void  free_group_id(struct group_statistics *v, void *aux UNUSED);
 
+	}
 
+      if ( v > 0 ) 
+	tab_hline(t, TAL_1, 0, n_cols - 1, (v * lines_per_variable) + 1);
+    }
 
+  tab_submit (t);
 
-int 
-compare_group_id (const struct group_statistics *a, 
-		  const struct group_statistics *b, int width)
-{
-  return compare_values(&a->id, &b->id, width);
 }
 
-unsigned int
-hash_group_id(const struct group_statistics *v, int width)
-{
-  return hash_value ( &v->id, width);
-}
 
-void 
-free_group_id(struct group_statistics *v, void *aux UNUSED)
-{
-  free(v);
-}
+/* ONEWAY ANOVA Calculations */
+
+static void  postcalc (  struct cmd_oneway *cmd UNUSED );
+
+static void  precalc ( struct cmd_oneway *cmd UNUSED );
+
 
 
 /* Pre calculations */
@@ -812,17 +893,17 @@ precalc ( struct cmd_oneway *cmd UNUSED )
 
   for(i=0; i< n_vars ; ++i) 
     {
-      struct group_statistics *totals = &vars[i]->p.ww.ugs;
+      struct group_statistics *totals = &vars[i]->p.grp_data.ugs;
       
       /* Create a hash for each of the dependent variables.
 	 The hash contains a group_statistics structure, 
 	 and is keyed by value of the independent variable */
 
-      vars[i]->p.ww.group_hash = 
+      vars[i]->p.grp_data.group_hash = 
 	hsh_create(4, 
-		   (hsh_compare_func *) compare_group_id,
-		   (hsh_hash_func *) hash_group_id,
-		   (hsh_free_func *) free_group_id,
+		   (hsh_compare_func *) compare_group,
+		   (hsh_hash_func *) hash_group,
+		   (hsh_free_func *) free_group,
 		   (void *) indep_var->width );
 
 
@@ -837,12 +918,11 @@ precalc ( struct cmd_oneway *cmd UNUSED )
 
 
 static void 
-calculate(const struct casefile *cf, void *cmd_)
+run_oneway(const struct casefile *cf, void *cmd_)
 {
   struct casereader *r;
   struct ccase c;
 
-
   struct cmd_oneway *cmd = (struct cmd_oneway *) cmd_;
 
   global_group_hash = hsh_create(4, 
@@ -850,7 +930,6 @@ calculate(const struct casefile *cf, void *cmd_)
 				 (hsh_hash_func *) hash_value,
 				 0,
 				 (void *) indep_var->width );
-
   precalc(cmd);
 
   for(r = casefile_get_reader (cf);
@@ -863,17 +942,37 @@ calculate(const struct casefile *cf, void *cmd_)
 	dict_get_case_weight(default_dict,&c,&bad_weight_warn);
       
       const union value *indep_val = case_data (&c, indep_var->fv);
+
+      /* Deal with missing values */
+      if ( value_is_missing(indep_val,indep_var) )
+	   continue;
+
+      /* Skip the entire case if /MISSING=LISTWISE is set */
+      if ( cmd->miss == ONEWAY_LISTWISE ) 
+	{
+	  for(i = 0; i < n_vars ; ++i) 
+	    {
+	      const struct variable *v = vars[i];
+	      const union value *val = case_data (&c, v->fv);
+
+	      if (value_is_missing(val,v) )
+		  break;
+	    }
+	  if ( i != n_vars ) 
+	      continue;
+
+	}
+      
 	  
       hsh_insert ( global_group_hash, (void *) indep_val );
 
-
       for ( i = 0 ; i < n_vars ; ++i ) 
 	{
 	  const struct variable *v = vars[i];
 
 	  const union value *val = case_data (&c, v->fv);
 
-	  struct hsh_table *group_hash = vars[i]->p.ww.group_hash;
+	  struct hsh_table *group_hash = vars[i]->p.grp_data.group_hash;
 
 	  struct group_statistics *gs;
 
@@ -897,7 +996,7 @@ calculate(const struct casefile *cf, void *cmd_)
 	  
 	  if (! value_is_missing(val,v) )
 	    {
-	      struct group_statistics *totals = &vars[i]->p.ww.ugs;
+	      struct group_statistics *totals = &vars[i]->p.grp_data.ugs;
 
 	      totals->n+=weight;
 	      totals->sum+=weight * val->f;
@@ -920,7 +1019,7 @@ calculate(const struct casefile *cf, void *cmd_)
 		gs->maximum = val->f * weight;
 	    }
 
-	  vars[i]->p.ww.n_groups = hsh_count ( group_hash );
+	  vars[i]->p.grp_data.n_groups = hsh_count ( group_hash );
 	}
   
     }
@@ -928,8 +1027,18 @@ calculate(const struct casefile *cf, void *cmd_)
 
   postcalc(cmd);
 
+  
+  if ( stat_tables & STAT_HOMO ) 
+    levene(cf, indep_var, n_vars, vars, 
+	   (cmd->miss == ONEWAY_LISTWISE) ? LEV_LISTWISE : LEV_ANALYSIS ,
+	   value_is_missing);
+
   ostensible_number_of_groups = hsh_count (global_group_hash);
 
+
+  output_oneway();
+
+
 }
 
 
@@ -942,8 +1051,8 @@ postcalc (  struct cmd_oneway *cmd UNUSED )
 
   for(i = 0; i < n_vars ; ++i) 
     {
-      struct hsh_table *group_hash = vars[i]->p.ww.group_hash;
-      struct group_statistics *totals = &vars[i]->p.ww.ugs;
+      struct hsh_table *group_hash = vars[i]->p.grp_data.group_hash;
+      struct group_statistics *totals = &vars[i]->p.grp_data.ugs;
 
       struct hsh_iterator g;
       struct group_statistics *gs;
@@ -977,8 +1086,5 @@ postcalc (  struct cmd_oneway *cmd UNUSED )
 
       totals->se_mean = totals->std_dev / sqrt(totals->n);
 	
-
-
-      
     }
 }