[Pspp-cvs] Changes to pspp/src/cat.c

[Jason H Stover]

Index: pspp/src/cat.c
diff -u pspp/src/cat.c:1.9 pspp/src/cat.c:1.10
--- pspp/src/cat.c:1.9  Wed Nov 23 19:40:37 2005
+++ pspp/src/cat.c      Sun Nov 27 20:25:51 2005
@@ -1,4 +1,4 @@
-/* PSPP - linear regression.
+/* PSPP - binary encodings for categorical variables.
    Copyright (C) 2005 Free Software Foundation, Inc.
    Written by Jason H Stover <address@hidden>.
 
@@ -36,18 +36,11 @@
 #include <error.h>
 #include "alloc.h"
 #include "error.h"
-#include "var.h"
 #include "cat.h"
+#include "cat-routines.h"
 #include <string.h>
-#include <math.h>
-#include <gsl/gsl_machine.h>
-#include <gsl/gsl_vector.h>
-#include <gsl/gsl_matrix.h>
 
 #define N_INITIAL_CATEGORIES 1
-#define CAT_COLUMN_NOT_FOUND -1
-#define CAT_VALUE_NOT_FOUND -2
-#define CAT_INDEX_NOT_FOUND -3
 
 void
 cat_stored_values_create (struct variable *v)
@@ -72,7 +65,10 @@
     }
 }
 
-static size_t
+/*
+  Which subscript corresponds to val?
+ */
+size_t
 cat_value_find (const struct variable *v, const union value *val)
 {
   size_t i;
@@ -121,35 +117,6 @@
     }
 }
 
-/*
-  Return the subscript of the binary vector corresponding
-  to this value.
- */
-size_t
-cat_value_to_subscript (const union value *val, struct variable *v)
-{
-  const union value *val2;
-  size_t subscript;
-  int different;
-
-  assert (v != NULL);
-  assert (val != NULL);
-  assert (v->obs_vals != NULL);
-  subscript = v->obs_vals->n_categories - 1;
-  while (subscript > 0)
-    {
-      val2 = v->obs_vals->vals + subscript;
-      assert (val2 != NULL);
-      different = compare_values (val, val2, v->width);
-      if (!different)
-       {
-         return subscript;
-       }
-      subscript--;
-    }
-  return subscript;
-}
-
 union value *
 cat_subscript_to_value (const size_t s, struct variable *v)
 {
@@ -162,237 +129,4 @@
     {
       return NULL;
     }
-}
-
-/*
-  Which element of a vector is equal to the value x?
- */
-static size_t
-cat_which_element_eq (const gsl_vector * vec, double x)
-{
-  size_t i;
-
-  for (i = 0; i < vec->size; i++)
-    {
-      if (fabs (gsl_vector_get (vec, i) - x) < GSL_DBL_EPSILON)
-       {
-         return i;
-       }
-    }
-  return CAT_VALUE_NOT_FOUND;
-}
-static int
-cat_is_zero_vector (const gsl_vector * vec)
-{
-  size_t i;
-
-  for (i = 0; i < vec->size; i++)
-    {
-      if (gsl_vector_get (vec, i) != 0.0)
-       {
-         return 0;
-       }
-    }
-  return 1;
-}
-
-/*
-  Return the value of v corresponding to the vector vec.
- */
-union value *
-cat_vector_to_value (const gsl_vector * vec, struct variable *v)
-{
-  size_t i;
-
-  i = cat_which_element_eq (vec, 1.0);
-  if (i != CAT_VALUE_NOT_FOUND)
-    {
-      return cat_subscript_to_value (i + 1, v);
-    }
-  if (cat_is_zero_vector (vec))
-    {
-      return cat_subscript_to_value (0, v);
-    }
-  return NULL;
-}
-
-struct design_matrix *
-design_matrix_create (int n_variables,
-                     const struct variable *v_variables[],
-                     const size_t n_data)
-{
-  struct design_matrix *dm;
-  const struct variable *v;
-  size_t i;
-  size_t n_cols = 0;
-  size_t col;
-
-  dm = xmalloc (sizeof *dm);
-  dm->vars = xnmalloc (n_variables, sizeof *dm->vars);
-  dm->n_vars = n_variables;
-
-  for (i = 0; i < n_variables; i++)
-    {
-      v = v_variables[i];
-      assert ((dm->vars + i) != NULL);
-      (dm->vars + i)->v = v;   /* Allows us to look up the variable from
-                                  the design matrix. */
-      (dm->vars + i)->first_column = n_cols;
-      if (v->type == NUMERIC)
-       {
-         n_cols++;
-         (dm->vars + i)->last_column = n_cols;
-       }
-      else if (v->type == ALPHA)
-       {
-         assert (v->obs_vals != NULL);
-         (dm->vars + i)->last_column =
-           (dm->vars + i)->first_column + v->obs_vals->n_categories - 2;
-         n_cols += v->obs_vals->n_categories - 1;
-       }
-    }
-  dm->m = gsl_matrix_calloc (n_data, n_cols);
-  col = 0;
-
-  return dm;
-}
-
-void
-design_matrix_destroy (struct design_matrix *dm)
-{
-  free (dm->vars);
-  gsl_matrix_free (dm->m);
-  free (dm);
-}
-
-/*
-  Return the index of the variable for the
-  given column.
- */
-static size_t
-design_matrix_col_to_var_index (const struct design_matrix *dm, size_t col)
-{
-  size_t i;
-  struct design_matrix_var v;
-
-  for (i = 0; i < dm->n_vars; i++)
-    {
-      v = dm->vars[i];
-      if (v.first_column <= col && col <= v.last_column)
-       return (v.v)->index;
-    }
-  return CAT_INDEX_NOT_FOUND;
-}
-
-/*
-  Return a pointer to the variable whose values
-  are stored in column col.
- */
-struct variable *
-design_matrix_col_to_var (const struct design_matrix *dm, size_t col)
-{
-  size_t index;
-  size_t i;
-  struct design_matrix_var dmv;
-
-  index = design_matrix_col_to_var_index (dm, col);
-  for (i = 0; i < dm->n_vars; i++)
-    {
-      dmv = dm->vars[i];
-      if ((dmv.v)->index == index)
-       {
-         return (struct variable *) dmv.v;
-       }
-    }
-  return NULL;
-}
-
-static size_t
-cmp_dm_var_index (const struct design_matrix_var *dmv, size_t index)
-{
-  if (dmv->v->index == index)
-    return 1;
-  return 0;
-}
-
-/*
-  Return the number of the first column which holds the
-  values for variable v.
- */
-size_t
-design_matrix_var_to_column (const struct design_matrix * dm,
-                            const struct variable * v)
-{
-  size_t i;
-  struct design_matrix_var tmp;
-
-  for (i = 0; i < dm->n_vars; i++)
-    {
-      tmp = dm->vars[i];
-      if (cmp_dm_var_index (&tmp, v->index))
-       {
-         return tmp.first_column;
-       }
-    }
-  return CAT_COLUMN_NOT_FOUND;
-}
-
-/* Last column. */
-static size_t
-dm_var_to_last_column (const struct design_matrix *dm,
-                      const struct variable *v)
-{
-  size_t i;
-  struct design_matrix_var tmp;
-
-  for (i = 0; i < dm->n_vars; i++)
-    {
-      tmp = dm->vars[i];
-      if (cmp_dm_var_index (&tmp, v->index))
-       {
-         return tmp.last_column;
-       }
-    }
-  return CAT_COLUMN_NOT_FOUND;
-}
-
-/*
-  Set the appropriate value in the design matrix, 
-  whether that value is from a categorical or numeric
-  variable. For a categorical variable, only the usual
-  binary encoding is allowed.
- */
-void
-design_matrix_set_categorical (struct design_matrix *dm, size_t row,
-                              const struct variable *var,
-                              const union value *val)
-{
-  size_t col;
-  size_t is_one;
-  size_t fc;
-  size_t lc;
-  double entry;
-
-  assert (var->type == ALPHA);
-  fc = design_matrix_var_to_column (dm, var);
-  lc = dm_var_to_last_column (dm, var);
-  assert (lc != CAT_COLUMN_NOT_FOUND);
-  assert (fc != CAT_COLUMN_NOT_FOUND);
-  is_one = fc + cat_value_find (var, val);
-  for (col = fc; col <= lc; col++)
-    {
-      entry = (col == is_one) ? 1.0 : 0.0;
-      gsl_matrix_set (dm->m, row, col, entry);
-    }
-}
-void
-design_matrix_set_numeric (struct design_matrix *dm, size_t row,
-                          const struct variable *var, const union value *val)
-{
-  size_t col;
-
-  assert (var->type == NUMERIC);
-  col = design_matrix_var_to_column ((const struct design_matrix *) dm, var);
-  assert (col != CAT_COLUMN_NOT_FOUND);
-  gsl_matrix_set (dm->m, row, col, val->f);
 }