X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fregression.q;h=0718b8d4d0bccce1da5ff3269ae0fd1b5ecc9157;hb=e385eeb8a2ea75fb2d9c1c628619baa03c914dae;hp=e38167900c4edfba0486c742fab18f8acdd6a22a;hpb=41a3a550334da96a9b4e5e089ad1768acf288092;p=pspp-builds.git

diff --git a/src/language/stats/regression.q b/src/language/stats/regression.q
index e3816790..0718b8d4 100644
--- a/src/language/stats/regression.q
+++ b/src/language/stats/regression.q
@@ -1,20 +1,18 @@
-/* PSPP - linear regression.
-   Copyright (C) 2005 Free Software Foundation, Inc.
+/* PSPP - a program for statistical analysis.
+   Copyright (C) 2005, 2009 Free Software Foundation, Inc.
 
-   This program is free software; you can redistribute it and/or
-   modify it under the terms of the GNU General Public License as
-   published by the Free Software Foundation; either version 2 of the
-   License, or (at your option) any later version.
+   This program is free software: you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation, either version 3 of the License, or
+   (at your option) any later version.
 
-   This program is distributed in the hope that it will be useful, but
-   WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-   General Public License for more details.
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
 
    You should have received a copy of the GNU General Public License
-   along with this program; if not, write to the Free Software
-   Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
-   02110-1301, USA. */
+   along with this program.  If not, see <http://www.gnu.org/licenses/>. */
 
 #include <config.h>
 
@@ -23,10 +21,9 @@
 #include <gsl/gsl_vector.h>
 #include <math.h>
 #include <stdlib.h>
-
-#include "regression-export.h"
 #include <data/case.h>
-#include <data/casefile.h>
+#include <data/casegrouper.h>
+#include <data/casereader.h>
 #include <data/category.h>
 #include <data/dictionary.h>
 #include <data/missing-values.h>
@@ -38,16 +35,18 @@
 #include <language/dictionary/split-file.h>
 #include <language/data-io/file-handle.h>
 #include <language/lexer/lexer.h>
-#include <libpspp/alloc.h>
 #include <libpspp/compiler.h>
 #include <libpspp/message.h>
-#include <math/design-matrix.h>
-#include <math/coefficient.h>
-#include <math/linreg/linreg.h>
+#include <libpspp/taint.h>
+#include <math/covariance.h>
+#include <math/linreg.h>
 #include <math/moments.h>
 #include <output/table.h>
 
+#include "xalloc.h"
+
 #include "gettext.h"
+#define _(msgid) gettext (msgid)
 
 #define REG_LARGE_DATA 1000
 
@@ -73,7 +72,6 @@
                     f,
                     defaults,
                     all;
-   export=custom;
    ^dependent=varlist;
    +save[sv_]=resid,pred;
    +method=enter.
@@ -91,9 +89,6 @@ struct moments_var
   const struct variable *v;
 };
 
-/* Linear regression models. */
-static pspp_linreg_cache **models = NULL;
-
 /*
   Transformations for saving predicted values
   and residuals, etc.
@@ -102,7 +97,7 @@ struct reg_trns
 {
   int n_trns;			/* Number of transformations. */
   int trns_id;			/* Which trns is this one? */
-  pspp_linreg_cache *c;		/* Linear model for this trns. */
+  linreg *c;		/* Linear model for this trns. */
 };
 /*
   Variables used (both explanatory and response).
@@ -114,44 +109,32 @@ static const struct variable **v_variables;
  */
 static size_t n_variables;
 
-/*
-  File where the model will be saved if the EXPORT subcommand
-  is given. 
- */
-static struct file_handle *model_file;
+static bool run_regression (struct casereader *, struct cmd_regression *,
+			    struct dataset *, linreg **);
 
 /*
-  Return value for the procedure.
- */
-static int pspp_reg_rc = CMD_SUCCESS;
-
-static bool run_regression (const struct ccase *,
-			    const struct casefile *, void *,
-			    const struct dataset *);
-
-/* 
    STATISTICS subcommand output functions.
  */
-static void reg_stats_r (pspp_linreg_cache *);
-static void reg_stats_coeff (pspp_linreg_cache *);
-static void reg_stats_anova (pspp_linreg_cache *);
-static void reg_stats_outs (pspp_linreg_cache *);
-static void reg_stats_zpp (pspp_linreg_cache *);
-static void reg_stats_label (pspp_linreg_cache *);
-static void reg_stats_sha (pspp_linreg_cache *);
-static void reg_stats_ci (pspp_linreg_cache *);
-static void reg_stats_f (pspp_linreg_cache *);
-static void reg_stats_bcov (pspp_linreg_cache *);
-static void reg_stats_ses (pspp_linreg_cache *);
-static void reg_stats_xtx (pspp_linreg_cache *);
-static void reg_stats_collin (pspp_linreg_cache *);
-static void reg_stats_tol (pspp_linreg_cache *);
-static void reg_stats_selection (pspp_linreg_cache *);
-static void statistics_keyword_output (void (*)(pspp_linreg_cache *),
-				       int, pspp_linreg_cache *);
+static void reg_stats_r (linreg *);
+static void reg_stats_coeff (linreg *);
+static void reg_stats_anova (linreg *);
+static void reg_stats_outs (linreg *);
+static void reg_stats_zpp (linreg *);
+static void reg_stats_label (linreg *);
+static void reg_stats_sha (linreg *);
+static void reg_stats_ci (linreg *);
+static void reg_stats_f (linreg *);
+static void reg_stats_bcov (linreg *);
+static void reg_stats_ses (linreg *);
+static void reg_stats_xtx (linreg *);
+static void reg_stats_collin (linreg *);
+static void reg_stats_tol (linreg *);
+static void reg_stats_selection (linreg *);
+static void statistics_keyword_output (void (*)(linreg *),
+				       int, linreg *);
 
 static void
-reg_stats_r (pspp_linreg_cache * c)
+reg_stats_r (linreg * c)
 {
   struct tab_table *t;
   int n_rows = 2;
@@ -161,11 +144,11 @@ reg_stats_r (pspp_linreg_cache * c)
   double std_error;
 
   assert (c != NULL);
-  rsq = c->ssm / c->sst;
-  adjrsq = 1.0 - (1.0 - rsq) * (c->n_obs - 1.0) / (c->n_obs - c->n_indeps);
-  std_error = sqrt ((c->n_indeps - 1.0) / (c->n_obs - 1.0));
+  rsq = linreg_ssreg (c) / linreg_sst (c);
+  adjrsq = 1.0 - (1.0 - rsq) * (linreg_n_obs (c) - 1.0) / (linreg_n_obs (c) - linreg_n_coeffs (c));
+  std_error = sqrt (linreg_mse (c));
   t = tab_create (n_cols, n_rows, 0);
-  tab_dim (t, tab_natural_dimensions);
+  tab_dim (t, tab_natural_dimensions, NULL);
   tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
   tab_hline (t, TAL_2, 0, n_cols - 1, 1);
   tab_vline (t, TAL_2, 2, 0, n_rows - 1);
@@ -175,10 +158,10 @@ reg_stats_r (pspp_linreg_cache * c)
   tab_text (t, 2, 0, TAB_CENTER | TAT_TITLE, _("R Square"));
   tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("Adjusted R Square"));
   tab_text (t, 4, 0, TAB_CENTER | TAT_TITLE, _("Std. Error of the Estimate"));
-  tab_float (t, 1, 1, TAB_RIGHT, sqrt (rsq), 10, 2);
-  tab_float (t, 2, 1, TAB_RIGHT, rsq, 10, 2);
-  tab_float (t, 3, 1, TAB_RIGHT, adjrsq, 10, 2);
-  tab_float (t, 4, 1, TAB_RIGHT, std_error, 10, 2);
+  tab_double (t, 1, 1, TAB_RIGHT, sqrt (rsq), NULL);
+  tab_double (t, 2, 1, TAB_RIGHT, rsq, NULL);
+  tab_double (t, 3, 1, TAB_RIGHT, adjrsq, NULL);
+  tab_double (t, 4, 1, TAB_RIGHT, std_error, NULL);
   tab_title (t, _("Model Summary"));
   tab_submit (t);
 }
@@ -187,30 +170,27 @@ reg_stats_r (pspp_linreg_cache * c)
   Table showing estimated regression coefficients.
  */
 static void
-reg_stats_coeff (pspp_linreg_cache * c)
+reg_stats_coeff (linreg * c)
 {
   size_t j;
   int n_cols = 7;
   int n_rows;
+  int this_row;
   double t_stat;
   double pval;
-  double coeff;
   double std_err;
   double beta;
   const char *label;
-  char *tmp;
+
   const struct variable *v;
-  const union value *val;
-  const char *val_s;
   struct tab_table *t;
 
   assert (c != NULL);
-  tmp = xnmalloc (MAX_STRING, sizeof (*tmp));
-  n_rows = c->n_coeffs + 2;
+  n_rows = linreg_n_coeffs (c) + 3;
 
   t = tab_create (n_cols, n_rows, 0);
   tab_headers (t, 2, 0, 1, 0);
-  tab_dim (t, tab_natural_dimensions);
+  tab_dim (t, tab_natural_dimensions, NULL);
   tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
   tab_hline (t, TAL_2, 0, n_cols - 1, 1);
   tab_vline (t, TAL_2, 2, 0, n_rows - 1);
@@ -222,82 +202,70 @@ reg_stats_coeff (pspp_linreg_cache * c)
   tab_text (t, 5, 0, TAB_CENTER | TAT_TITLE, _("t"));
   tab_text (t, 6, 0, TAB_CENTER | TAT_TITLE, _("Significance"));
   tab_text (t, 1, 1, TAB_LEFT | TAT_TITLE, _("(Constant)"));
-  coeff = c->coeff[0]->estimate;
-  tab_float (t, 2, 1, 0, coeff, 10, 2);
-  std_err = sqrt (gsl_matrix_get (c->cov, 0, 0));
-  tab_float (t, 3, 1, 0, std_err, 10, 2);
-  beta = coeff / c->depvar_std;
-  tab_float (t, 4, 1, 0, beta, 10, 2);
-  t_stat = coeff / std_err;
-  tab_float (t, 5, 1, 0, t_stat, 10, 2);
+  tab_double (t, 2, 1, 0, linreg_intercept (c), NULL);
+  std_err = sqrt (gsl_matrix_get (linreg_cov (c), 0, 0));
+  tab_double (t, 3, 1, 0, std_err, NULL);
+  tab_double (t, 4, 1, 0, 0.0, NULL);
+  t_stat = linreg_intercept (c) / std_err;
+  tab_double (t, 5, 1, 0, t_stat, NULL);
   pval = 2 * gsl_cdf_tdist_Q (fabs (t_stat), 1.0);
-  tab_float (t, 6, 1, 0, pval, 10, 2);
-  for (j = 1; j <= c->n_indeps; j++)
+  tab_double (t, 6, 1, 0, pval, NULL);
+  for (j = 0; j < linreg_n_coeffs (c); j++)
     {
-      v = pspp_coeff_get_var (c->coeff[j], 0);
+      struct string tstr;
+      ds_init_empty (&tstr);
+      this_row = j + 2;
+
+      v = linreg_indep_var (c, j);
       label = var_to_string (v);
       /* Do not overwrite the variable's name. */
-      strncpy (tmp, label, MAX_STRING);
-      if (var_is_alpha (v))
-	{
-	  /*
-	     Append the value associated with this coefficient.
-	     This makes sense only if we us the usual binary encoding
-	     for that value.
-	   */
-
-	  val = pspp_coeff_get_value (c->coeff[j], v);
-	  val_s = var_get_value_name (v, val);
-	  strncat (tmp, val_s, MAX_STRING);
-	}
-
-      tab_text (t, 1, j + 1, TAB_CENTER, tmp);
+      ds_put_cstr (&tstr, label);
+      tab_text (t, 1, this_row, TAB_CENTER, ds_cstr (&tstr));
       /*
          Regression coefficients.
        */
-      coeff = c->coeff[j]->estimate;
-      tab_float (t, 2, j + 1, 0, coeff, 10, 2);
+      tab_double (t, 2, this_row, 0, linreg_coeff (c, j), NULL);
       /*
          Standard error of the coefficients.
        */
-      std_err = sqrt (gsl_matrix_get (c->cov, j, j));
-      tab_float (t, 3, j + 1, 0, std_err, 10, 2);
+      std_err = sqrt (gsl_matrix_get (linreg_cov (c), j + 1, j + 1));
+      tab_double (t, 3, this_row, 0, std_err, NULL);
       /*
-         'Standardized' coefficient, i.e., regression coefficient
+         Standardized coefficient, i.e., regression coefficient
          if all variables had unit variance.
        */
-      beta = gsl_vector_get (c->indep_std, j);
-      beta *= coeff / c->depvar_std;
-      tab_float (t, 4, j + 1, 0, beta, 10, 2);
+      beta = sqrt (gsl_matrix_get (linreg_cov (c), j, j));
+      beta *= linreg_coeff (c, j) / c->depvar_std;
+      tab_double (t, 4, this_row, 0, beta, NULL);
 
       /*
          Test statistic for H0: coefficient is 0.
        */
-      t_stat = coeff / std_err;
-      tab_float (t, 5, j + 1, 0, t_stat, 10, 2);
+      t_stat = linreg_coeff (c, j) / std_err;
+      tab_double (t, 5, this_row, 0, t_stat, NULL);
       /*
          P values for the test statistic above.
        */
       pval =
 	2 * gsl_cdf_tdist_Q (fabs (t_stat),
-			     (double) (c->n_obs - c->n_coeffs));
-      tab_float (t, 6, j + 1, 0, pval, 10, 2);
+			     (double) (linreg_n_obs (c) - linreg_n_coeffs (c)));
+      tab_double (t, 6, this_row, 0, pval, NULL);
+      ds_destroy (&tstr);
     }
   tab_title (t, _("Coefficients"));
   tab_submit (t);
-  free (tmp);
 }
 
 /*
   Display the ANOVA table.
  */
 static void
-reg_stats_anova (pspp_linreg_cache * c)
+reg_stats_anova (linreg * c)
 {
   int n_cols = 7;
   int n_rows = 4;
-  const double msm = c->ssm / c->dfm;
-  const double mse = c->sse / c->dfe;
+  const double msm = linreg_ssreg (c) / linreg_dfmodel (c);
+  const double mse = linreg_mse (c);
   const double F = msm / mse;
   const double pval = gsl_cdf_fdist_Q (F, c->dfm, c->dfe);
 
@@ -306,7 +274,7 @@ reg_stats_anova (pspp_linreg_cache * c)
   assert (c != NULL);
   t = tab_create (n_cols, n_rows, 0);
   tab_headers (t, 2, 0, 1, 0);
-  tab_dim (t, tab_natural_dimensions);
+  tab_dim (t, tab_natural_dimensions, NULL);
 
   tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
 
@@ -325,60 +293,63 @@ reg_stats_anova (pspp_linreg_cache * c)
   tab_text (t, 1, 3, TAB_LEFT | TAT_TITLE, _("Total"));
 
   /* Sums of Squares */
-  tab_float (t, 2, 1, 0, c->ssm, 10, 2);
-  tab_float (t, 2, 3, 0, c->sst, 10, 2);
-  tab_float (t, 2, 2, 0, c->sse, 10, 2);
+  tab_double (t, 2, 1, 0, linreg_ssreg (c), NULL);
+  tab_double (t, 2, 3, 0, linreg_sst (c), NULL);
+  tab_double (t, 2, 2, 0, linreg_sse (c), NULL);
 
 
   /* Degrees of freedom */
-  tab_float (t, 3, 1, 0, c->dfm, 4, 0);
-  tab_float (t, 3, 2, 0, c->dfe, 4, 0);
-  tab_float (t, 3, 3, 0, c->dft, 4, 0);
+  tab_text_format (t, 3, 1, TAB_RIGHT, "%g", c->dfm);
+  tab_text_format (t, 3, 2, TAB_RIGHT, "%g", c->dfe);
+  tab_text_format (t, 3, 3, TAB_RIGHT, "%g", c->dft);
 
   /* Mean Squares */
+  tab_double (t, 4, 1, TAB_RIGHT, msm, NULL);
+  tab_double (t, 4, 2, TAB_RIGHT, mse, NULL);
 
-  tab_float (t, 4, 1, TAB_RIGHT, msm, 8, 3);
-  tab_float (t, 4, 2, TAB_RIGHT, mse, 8, 3);
-
-  tab_float (t, 5, 1, 0, F, 8, 3);
+  tab_double (t, 5, 1, 0, F, NULL);
 
-  tab_float (t, 6, 1, 0, pval, 8, 3);
+  tab_double (t, 6, 1, 0, pval, NULL);
 
   tab_title (t, _("ANOVA"));
   tab_submit (t);
 }
+
 static void
-reg_stats_outs (pspp_linreg_cache * c)
+reg_stats_outs (linreg * c)
 {
   assert (c != NULL);
 }
+
 static void
-reg_stats_zpp (pspp_linreg_cache * c)
+reg_stats_zpp (linreg * c)
 {
   assert (c != NULL);
 }
+
 static void
-reg_stats_label (pspp_linreg_cache * c)
+reg_stats_label (linreg * c)
 {
   assert (c != NULL);
 }
+
 static void
-reg_stats_sha (pspp_linreg_cache * c)
+reg_stats_sha (linreg * c)
 {
   assert (c != NULL);
 }
 static void
-reg_stats_ci (pspp_linreg_cache * c)
+reg_stats_ci (linreg * c)
 {
   assert (c != NULL);
 }
 static void
-reg_stats_f (pspp_linreg_cache * c)
+reg_stats_f (linreg * c)
 {
   assert (c != NULL);
 }
 static void
-reg_stats_bcov (pspp_linreg_cache * c)
+reg_stats_bcov (linreg * c)
 {
   int n_cols;
   int n_rows;
@@ -394,59 +365,59 @@ reg_stats_bcov (pspp_linreg_cache * c)
   n_rows = 2 * (c->n_indeps + 1);
   t = tab_create (n_cols, n_rows, 0);
   tab_headers (t, 2, 0, 1, 0);
-  tab_dim (t, tab_natural_dimensions);
+  tab_dim (t, tab_natural_dimensions, NULL);
   tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
   tab_hline (t, TAL_2, 0, n_cols - 1, 1);
   tab_vline (t, TAL_2, 2, 0, n_rows - 1);
   tab_vline (t, TAL_0, 1, 0, 0);
   tab_text (t, 0, 0, TAB_CENTER | TAT_TITLE, _("Model"));
   tab_text (t, 1, 1, TAB_CENTER | TAT_TITLE, _("Covariances"));
-  for (i = 1; i < c->n_coeffs; i++)
+  for (i = 0; i < linreg_n_coeffs (c); i++)
     {
-      const struct variable *v = pspp_coeff_get_var (c->coeff[i], 0);
+      const struct variable *v = linreg_indep_var (c, i);
       label = var_to_string (v);
       tab_text (t, 2, i, TAB_CENTER, label);
       tab_text (t, i + 2, 0, TAB_CENTER, label);
-      for (k = 1; k < c->n_coeffs; k++)
+      for (k = 1; k < linreg_n_coeffs (c); k++)
 	{
 	  col = (i <= k) ? k : i;
 	  row = (i <= k) ? i : k;
-	  tab_float (t, k + 2, i, TAB_CENTER,
-		     gsl_matrix_get (c->cov, row, col), 8, 3);
+	  tab_double (t, k + 2, i, TAB_CENTER,
+		     gsl_matrix_get (c->cov, row, col), NULL);
 	}
     }
   tab_title (t, _("Coefficient Correlations"));
   tab_submit (t);
 }
 static void
-reg_stats_ses (pspp_linreg_cache * c)
+reg_stats_ses (linreg * c)
 {
   assert (c != NULL);
 }
 static void
-reg_stats_xtx (pspp_linreg_cache * c)
+reg_stats_xtx (linreg * c)
 {
   assert (c != NULL);
 }
 static void
-reg_stats_collin (pspp_linreg_cache * c)
+reg_stats_collin (linreg * c)
 {
   assert (c != NULL);
 }
 static void
-reg_stats_tol (pspp_linreg_cache * c)
+reg_stats_tol (linreg * c)
 {
   assert (c != NULL);
 }
 static void
-reg_stats_selection (pspp_linreg_cache * c)
+reg_stats_selection (linreg * c)
 {
   assert (c != NULL);
 }
 
 static void
-statistics_keyword_output (void (*function) (pspp_linreg_cache *),
-			   int keyword, pspp_linreg_cache * c)
+statistics_keyword_output (void (*function) (linreg *),
+			   int keyword, linreg * c)
 {
   if (keyword)
     {
@@ -455,10 +426,10 @@ statistics_keyword_output (void (*function) (pspp_linreg_cache *),
 }
 
 static void
-subcommand_statistics (int *keywords, pspp_linreg_cache * c)
+subcommand_statistics (int *keywords, linreg * c)
 {
-  /* 
-     The order here must match the order in which the STATISTICS 
+  /*
+     The order here must match the order in which the STATISTICS
      keywords appear in the specification section above.
    */
   enum
@@ -544,7 +515,7 @@ regression_trns_free (void *t_)
 
   if (t->trns_id == t->n_trns)
     {
-      result = pspp_linreg_cache_free (t->c);
+      result = linreg_free (t->c);
     }
   free (t);
 
@@ -555,15 +526,16 @@ regression_trns_free (void *t_)
   Gets the predicted values.
  */
 static int
-regression_trns_pred_proc (void *t_, struct ccase *c,
+regression_trns_pred_proc (void *t_, struct ccase **c,
 			   casenumber case_idx UNUSED)
 {
   size_t i;
   size_t n_vals;
   struct reg_trns *trns = t_;
-  pspp_linreg_cache *model;
+  linreg *model;
   union value *output = NULL;
-  const union value **vals = NULL;
+  const union value *tmp;
+  double *vals;
   const struct variable **vars = NULL;
 
   assert (trns != NULL);
@@ -572,21 +544,20 @@ regression_trns_pred_proc (void *t_, struct ccase *c,
   assert (model->depvar != NULL);
   assert (model->pred != NULL);
 
-  vars = xnmalloc (model->n_coeffs, sizeof (*vars));
-  n_vals = (*model->get_vars) (model, vars);
-
+  vars = linreg_get_vars (model);
+  n_vals = linreg_n_coeffs (model);
   vals = xnmalloc (n_vals, sizeof (*vals));
-  output = case_data_rw (c, model->pred);
-  assert (output != NULL);
+  *c = case_unshare (*c);
+
+  output = case_data_rw (*c, model->pred);
 
   for (i = 0; i < n_vals; i++)
     {
-      vals[i] = case_data (c, vars[i]);
+      tmp = case_data (*c, vars[i]);
+      vals[i] = tmp->f;
     }
-  output->f = (*model->predict) ((const struct variable **) vars,
-				 vals, model, n_vals);
+  output->f = linreg_predict (model, vals, n_vals);
   free (vals);
-  free (vars);
   return TRNS_CONTINUE;
 }
 
@@ -594,16 +565,17 @@ regression_trns_pred_proc (void *t_, struct ccase *c,
   Gets the residuals.
  */
 static int
-regression_trns_resid_proc (void *t_, struct ccase *c,
+regression_trns_resid_proc (void *t_, struct ccase **c,
 			    casenumber case_idx UNUSED)
 {
   size_t i;
   size_t n_vals;
   struct reg_trns *trns = t_;
-  pspp_linreg_cache *model;
+  linreg *model;
   union value *output = NULL;
-  const union value **vals = NULL;
-  const union value *obs = NULL;
+  const union value *tmp;
+  double *vals = NULL;
+  double obs;
   const struct variable **vars = NULL;
 
   assert (trns != NULL);
@@ -612,26 +584,28 @@ regression_trns_resid_proc (void *t_, struct ccase *c,
   assert (model->depvar != NULL);
   assert (model->resid != NULL);
 
-  vars = xnmalloc (model->n_coeffs, sizeof (*vars));
-  n_vals = (*model->get_vars) (model, vars);
+  vars = linreg_get_vars (model);
+  n_vals = linreg_n_coeffs (model);
 
   vals = xnmalloc (n_vals, sizeof (*vals));
-  output = case_data_rw (c, model->resid);
+  *c = case_unshare (*c);
+  output = case_data_rw (*c, model->resid);
   assert (output != NULL);
 
   for (i = 0; i < n_vals; i++)
     {
-      vals[i] = case_data (c, vars[i]);
+      tmp = case_data (*c, vars[i]);
+      vals[i] = tmp->f;
     }
-  obs = case_data (c, model->depvar);
-  output->f = (*model->residual) ((const struct variable **) vars,
-				  vals, obs, model, n_vals);
+  tmp = case_data (*c, model->depvar);
+  obs = tmp->f;
+  output->f = linreg_residual (model, obs, vals, n_vals);
   free (vals);
-  free (vars);
+
   return TRNS_CONTINUE;
 }
 
-/* 
+/*
    Returns false if NAME is a duplicate of any existing variable name.
 */
 static bool
@@ -644,26 +618,26 @@ try_name (const struct dictionary *dict, const char *name)
 }
 
 static void
-reg_get_name (const struct dictionary *dict, char name[LONG_NAME_LEN],
-	      const char prefix[LONG_NAME_LEN])
+reg_get_name (const struct dictionary *dict, char name[VAR_NAME_LEN],
+	      const char prefix[VAR_NAME_LEN])
 {
   int i = 1;
 
-  snprintf (name, LONG_NAME_LEN, "%s%d", prefix, i);
+  snprintf (name, VAR_NAME_LEN, "%s%d", prefix, i);
   while (!try_name (dict, name))
     {
       i++;
-      snprintf (name, LONG_NAME_LEN, "%s%d", prefix, i);
+      snprintf (name, VAR_NAME_LEN, "%s%d", prefix, i);
     }
 }
 
 static void
 reg_save_var (struct dataset *ds, const char *prefix, trns_proc_func * f,
-	      pspp_linreg_cache * c, struct variable **v, int n_trns)
+	      linreg * c, struct variable **v, int n_trns)
 {
   struct dictionary *dict = dataset_dict (ds);
   static int trns_index = 1;
-  char name[LONG_NAME_LEN];
+  char name[VAR_NAME_LEN];
   struct variable *new_var;
   struct reg_trns *t = NULL;
 
@@ -678,16 +652,13 @@ reg_save_var (struct dataset *ds, const char *prefix, trns_proc_func * f,
   add_transformation (ds, f, regression_trns_free, t);
   trns_index++;
 }
-
 static void
-subcommand_save (struct dataset *ds, int save, pspp_linreg_cache ** models)
+subcommand_save (struct dataset *ds, int save, linreg ** models)
 {
-  pspp_linreg_cache **lc;
+  linreg **lc;
   int n_trns = 0;
   int i;
 
-  assert (models != NULL);
-
   if (save)
     {
       /* Count the number of transformations we will need. */
@@ -702,17 +673,21 @@ subcommand_save (struct dataset *ds, int save, pspp_linreg_cache ** models)
 
       for (lc = models; lc < models + cmd.n_dependent; lc++)
 	{
-	  assert (*lc != NULL);
-	  assert ((*lc)->depvar != NULL);
-	  if (cmd.a_save[REGRESSION_SV_RESID])
-	    {
-	      reg_save_var (ds, "RES", regression_trns_resid_proc, *lc,
-			    &(*lc)->resid, n_trns);
-	    }
-	  if (cmd.a_save[REGRESSION_SV_PRED])
+	  if (*lc != NULL)
 	    {
-	      reg_save_var (ds, "PRED", regression_trns_pred_proc, *lc,
-			    &(*lc)->pred, n_trns);
+	      if ((*lc)->depvar != NULL)
+		{
+		  if (cmd.a_save[REGRESSION_SV_RESID])
+		    {
+		      reg_save_var (ds, "RES", regression_trns_resid_proc, *lc,
+				    &(*lc)->resid, n_trns);
+		    }
+		  if (cmd.a_save[REGRESSION_SV_PRED])
+		    {
+		      reg_save_var (ds, "PRED", regression_trns_pred_proc, *lc,
+				    &(*lc)->pred, n_trns);
+		    }
+		}
 	    }
 	}
     }
@@ -722,251 +697,45 @@ subcommand_save (struct dataset *ds, int save, pspp_linreg_cache ** models)
 	{
 	  if (*lc != NULL)
 	    {
-	      pspp_linreg_cache_free (*lc);
-	    }
-	}
-    }
-}
-
-static int
-reg_inserted (const struct variable *v, struct variable **varlist, int n_vars)
-{
-  int i;
-
-  for (i = 0; i < n_vars; i++)
-    {
-      if (v == varlist[i])
-	{
-	  return 1;
-	}
-    }
-  return 0;
-}
-
-static void
-reg_print_categorical_encoding (FILE * fp, pspp_linreg_cache * c)
-{
-  int i;
-  int n_vars = 0;
-  struct variable **varlist;
-
-  fprintf (fp, "%s", reg_export_categorical_encode_1);
-
-  varlist = xnmalloc (c->n_indeps, sizeof (*varlist));
-  for (i = 1; i < c->n_indeps; i++)	/* c->coeff[0] is the intercept. */
-    {
-      struct pspp_coeff *coeff = c->coeff[i];
-      const struct variable *v = pspp_coeff_get_var (coeff, 0);
-      if (var_is_alpha (v))
-	{
-	  if (!reg_inserted (v, varlist, n_vars))
-	    {
-	      fprintf (fp, "struct pspp_reg_categorical_variable %s;\n\t",
-		       var_get_name (v));
-	      varlist[n_vars] = (struct variable *) v;
-	      n_vars++;
-	    }
-	}
-    }
-  fprintf (fp, "int n_vars = %d;\n\t", n_vars);
-  fprintf (fp, "struct pspp_reg_categorical_variable *varlist[%d] = {",
-	   n_vars);
-  for (i = 0; i < n_vars - 1; i++)
-    {
-      fprintf (fp, "&%s,\n\t\t", var_get_name (varlist[i]));
-    }
-  fprintf (fp, "&%s};\n\t", var_get_name (varlist[i]));
-
-  for (i = 0; i < n_vars; i++)
-    {
-      int n_categories = cat_get_n_categories (varlist[i]);
-      int j;
-
-      fprintf (fp, "%s.name = \"%s\";\n\t",
-	       var_get_name (varlist[i]), var_get_name (varlist[i]));
-      fprintf (fp, "%s.n_vals = %d;\n\t",
-	       var_get_name (varlist[i]), n_categories);
-
-      for (j = 0; j < n_categories; j++)
-	{
-	  const union value *val = cat_subscript_to_value (j, varlist[i]);
-	  fprintf (fp, "%s.values[%d] = \"%s\";\n\t",
-		   var_get_name (varlist[i]), j,
-		   var_get_value_name (varlist[i], val));
-	}
-    }
-  fprintf (fp, "%s", reg_export_categorical_encode_2);
-}
-
-static void
-reg_print_depvars (FILE * fp, pspp_linreg_cache * c)
-{
-  int i;
-  struct pspp_coeff *coeff;
-  const struct variable *v;
-
-  fprintf (fp, "char *model_depvars[%d] = {", c->n_indeps);
-  for (i = 1; i < c->n_indeps; i++)
-    {
-      coeff = c->coeff[i];
-      v = pspp_coeff_get_var (coeff, 0);
-      fprintf (fp, "\"%s\",\n\t\t", var_get_name (v));
-    }
-  coeff = c->coeff[i];
-  v = pspp_coeff_get_var (coeff, 0);
-  fprintf (fp, "\"%s\"};\n\t", var_get_name (v));
-}
-static void
-reg_print_getvar (FILE * fp, pspp_linreg_cache * c)
-{
-  fprintf (fp, "static int\npspp_reg_getvar (char *v_name)\n{\n\t");
-  fprintf (fp, "int i;\n\tint n_vars = %d;\n\t", c->n_indeps);
-  reg_print_depvars (fp, c);
-  fprintf (fp, "for (i = 0; i < n_vars; i++)\n\t{\n\t\t");
-  fprintf (fp,
-	   "if (strncmp (v_name, model_depvars[i], PSPP_REG_MAXLEN) == 0)\n\t\t{\n\t\t\t");
-  fprintf (fp, "return i;\n\t\t}\n\t}\n}\n");
-}
-static int
-reg_has_categorical (pspp_linreg_cache * c)
-{
-  int i;
-  const struct variable *v;
-
-  for (i = 1; i < c->n_coeffs; i++)
-    {
-      v = pspp_coeff_get_var (c->coeff[i], 0);
-      if (var_is_alpha (v))
-	return 1;
-    }
-  return 0;
-}
-
-static void
-subcommand_export (int export, pspp_linreg_cache * c)
-{
-  FILE *fp;
-  size_t i;
-  size_t j;
-  int n_quantiles = 100;
-  double tmp;
-  struct pspp_coeff *coeff;
-
-  if (export)
-    {
-      assert (c != NULL);
-      assert (model_file != NULL);
-      fp = fopen (fh_get_file_name (model_file), "w");
-      assert (fp != NULL);
-      fprintf (fp, "%s", reg_preamble);
-      reg_print_getvar (fp, c);
-      if (reg_has_categorical (c))
-	{
-	  reg_print_categorical_encoding (fp, c);
-	}
-      fprintf (fp, "%s", reg_export_t_quantiles_1);
-      for (i = 0; i < n_quantiles - 1; i++)
-	{
-	  tmp = 0.5 + 0.005 * (double) i;
-	  fprintf (fp, "%.15e,\n\t\t",
-		   gsl_cdf_tdist_Pinv (tmp, c->n_obs - c->n_indeps));
-	}
-      fprintf (fp, "%.15e};\n\t",
-	       gsl_cdf_tdist_Pinv (.9995, c->n_obs - c->n_indeps));
-      fprintf (fp, "%s", reg_export_t_quantiles_2);
-      fprintf (fp, "%s", reg_mean_cmt);
-      fprintf (fp, "double\npspp_reg_estimate (const double *var_vals,");
-      fprintf (fp, "const char *var_names[])\n{\n\t");
-      fprintf (fp, "double model_coeffs[%d] = {", c->n_indeps);
-      for (i = 1; i < c->n_indeps; i++)
-	{
-	  coeff = c->coeff[i];
-	  fprintf (fp, "%.15e,\n\t\t", coeff->estimate);
-	}
-      coeff = c->coeff[i];
-      fprintf (fp, "%.15e};\n\t", coeff->estimate);
-      coeff = c->coeff[0];
-      fprintf (fp, "double estimate = %.15e;\n\t", coeff->estimate);
-      fprintf (fp, "int i;\n\tint j;\n\n\t");
-      fprintf (fp, "for (i = 0; i < %d; i++)\n\t", c->n_indeps);
-      fprintf (fp, "%s", reg_getvar);
-      fprintf (fp, "const double cov[%d][%d] = {\n\t", c->n_coeffs,
-	       c->n_coeffs);
-      for (i = 0; i < c->cov->size1 - 1; i++)
-	{
-	  fprintf (fp, "{");
-	  for (j = 0; j < c->cov->size2 - 1; j++)
-	    {
-	      fprintf (fp, "%.15e, ", gsl_matrix_get (c->cov, i, j));
+	      linreg_free (*lc);
 	    }
-	  fprintf (fp, "%.15e},\n\t", gsl_matrix_get (c->cov, i, j));
-	}
-      fprintf (fp, "{");
-      for (j = 0; j < c->cov->size2 - 1; j++)
-	{
-	  fprintf (fp, "%.15e, ",
-		   gsl_matrix_get (c->cov, c->cov->size1 - 1, j));
 	}
-      fprintf (fp, "%.15e}\n\t",
-	       gsl_matrix_get (c->cov, c->cov->size1 - 1, c->cov->size2 - 1));
-      fprintf (fp, "};\n\tint n_vars = %d;\n\tint i;\n\tint j;\n\t",
-	       c->n_indeps);
-      fprintf (fp, "double unshuffled_vals[%d];\n\t", c->n_indeps);
-      fprintf (fp, "%s", reg_variance);
-      fprintf (fp, "%s", reg_export_confidence_interval);
-      tmp = c->mse * c->mse;
-      fprintf (fp, "%s %.15e", reg_export_prediction_interval_1, tmp);
-      fprintf (fp, "%s %.15e", reg_export_prediction_interval_2, tmp);
-      fprintf (fp, "%s", reg_export_prediction_interval_3);
-      fclose (fp);
-      fp = fopen ("pspp_model_reg.h", "w");
-      fprintf (fp, "%s", reg_header);
-      fclose (fp);
     }
 }
 
-static int
-regression_custom_export (struct lexer *lexer, struct dataset *ds UNUSED,
-			  struct cmd_regression *cmd UNUSED, void *aux UNUSED)
-{
-  /* 0 on failure, 1 on success, 2 on failure that should result in syntax error */
-  if (!lex_force_match (lexer, '('))
-    return 0;
-
-  if (lex_match (lexer, '*'))
-    model_file = NULL;
-  else
-    {
-      model_file = fh_parse (lexer, FH_REF_FILE);
-      if (model_file == NULL)
-	return 0;
-    }
-
-  if (!lex_force_match (lexer, ')'))
-    return 0;
-
-  return 1;
-}
-
 int
 cmd_regression (struct lexer *lexer, struct dataset *ds)
 {
+  struct casegrouper *grouper;
+  struct casereader *group;
+  linreg **models;
+  bool ok;
   size_t i;
 
   if (!parse_regression (lexer, ds, &cmd, NULL))
-    return CMD_FAILURE;
+    {
+      return CMD_FAILURE;
+    }
 
   models = xnmalloc (cmd.n_dependent, sizeof *models);
   for (i = 0; i < cmd.n_dependent; i++)
     {
       models[i] = NULL;
     }
-  if (!multipass_procedure_with_splits (ds, run_regression, &cmd))
-    return CMD_CASCADING_FAILURE;
+
+  /* Data pass. */
+  grouper = casegrouper_create_splits (proc_open (ds), dataset_dict (ds));
+  while (casegrouper_get_next_group (grouper, &group))
+    run_regression (group, &cmd, ds, models);
+  ok = casegrouper_destroy (grouper);
+  ok = proc_commit (ds) && ok;
+
   subcommand_save (ds, cmd.sbc_save, models);
   free (v_variables);
   free (models);
-  return pspp_reg_rc;
+  free_regression (&cmd);
+
+  return ok ? CMD_SUCCESS : CMD_FAILURE;
 }
 
 /*
@@ -978,47 +747,6 @@ is_depvar (size_t k, const struct variable *v)
   return v == v_variables[k];
 }
 
-/*
-  Mark missing cases. Return the number of non-missing cases.
-  Compute the first two moments.
- */
-static size_t
-mark_missing_cases (const struct casefile *cf, const struct variable *v,
-		    int *is_missing_case, double n_data,
-		    struct moments_var *mom)
-{
-  struct casereader *r;
-  struct ccase c;
-  size_t row;
-  const union value *val;
-  double w = 1.0;
-
-  for (r = casefile_get_reader (cf, NULL);
-       casereader_read (r, &c); case_destroy (&c))
-    {
-      row = casereader_cnum (r) - 1;
-
-      val = case_data (&c, v);
-      if (mom != NULL)
-	{
-	  moments1_add (mom->m, val->f, w);
-	}
-      cat_value_update (v, val);
-      if (var_is_value_missing (v, val, MV_ANY))
-	{
-	  if (!is_missing_case[row])
-	    {
-	      /* Now it is missing. */
-	      n_data--;
-	      is_missing_case[row] = 1;
-	    }
-	}
-    }
-  casereader_destroy (r);
-
-  return n_data;
-}
-
 /* Parser for the variables sub command */
 static int
 regression_custom_variables (struct lexer *lexer, struct dataset *ds,
@@ -1046,281 +774,206 @@ regression_custom_variables (struct lexer *lexer, struct dataset *ds,
   return 1;
 }
 
-/*
-  Count the explanatory variables. The user may or may
-  not have specified a response variable in the syntax.
- */
-static int
-get_n_indep (const struct variable *v)
-{
-  int result;
-  int i = 0;
-
-  result = n_variables;
-  while (i < n_variables)
-    {
-      if (is_depvar (i, v))
-	{
-	  result--;
-	  i = n_variables;
-	}
-      i++;
-    }
-  return result;
-}
-
-/*
-  Read from the active file. Identify the explanatory variables in
-  v_variables. Encode categorical variables. Drop cases with missing
-  values.
-*/
+/* Identify the explanatory variables in v_variables.  Returns
+   the number of independent variables. */
 static int
-prepare_data (int n_data, int is_missing_case[],
-	      const struct variable **indep_vars,
-	      const struct variable *depvar, const struct casefile *cf,
-	      struct moments_var *mom)
+identify_indep_vars (const struct variable **indep_vars,
+		     const struct variable *depvar)
 {
+  int n_indep_vars = 0;
   int i;
-  int j;
 
-  assert (indep_vars != NULL);
-  j = 0;
   for (i = 0; i < n_variables; i++)
+    if (!is_depvar (i, depvar))
+      indep_vars[n_indep_vars++] = v_variables[i];
+  if ((n_indep_vars < 1) && is_depvar (0, depvar))
     {
-      if (!is_depvar (i, depvar))
-	{
-	  indep_vars[j] = v_variables[i];
-	  j++;
-	  if (var_is_alpha (v_variables[i]))
-	    {
-	      /* Make a place to hold the binary vectors
-	         corresponding to this variable's values. */
-	      cat_stored_values_create (v_variables[i]);
-	    }
-	  n_data =
-	    mark_missing_cases (cf, v_variables[i], is_missing_case, n_data,
-				mom + i);
-	}
+      /*
+	There is only one independent variable, and it is the same
+	as the dependent variable. Print a warning and continue.
+       */
+      msg (SE,
+	   gettext ("The dependent variable is equal to the independent variable." 
+		    "The least squares line is therefore Y=X." 
+		    "Standard errors and related statistics may be meaningless."));
+      n_indep_vars = 1;
+      indep_vars[0] = v_variables[0];
     }
-  /*
-     Mark missing cases for the dependent variable.
-   */
-  n_data = mark_missing_cases (cf, depvar, is_missing_case, n_data, NULL);
-
-  return n_data;
+  return n_indep_vars;
 }
-static void
-coeff_init (pspp_linreg_cache * c, struct design_matrix *dm)
-{
-  c->coeff = xnmalloc (dm->m->size2 + 1, sizeof (*c->coeff));
-  c->coeff[0] = xmalloc (sizeof (*(c->coeff[0])));	/* The first coefficient is the intercept. */
-  c->coeff[0]->v_info = NULL;	/* Intercept has no associated variable. */
-  pspp_coeff_init (c->coeff + 1, dm);
-}
-
-/*
-  Put the moments in the linreg cache.
- */
-static void
-compute_moments (pspp_linreg_cache * c, struct moments_var *mom,
-		 struct design_matrix *dm, size_t n)
+static double
+fill_covariance (gsl_matrix *cov, struct covariance *all_cov, 
+		 const struct variable **vars,
+		 size_t n_vars, const struct variable *dep_var, 
+		 const struct variable **all_vars, size_t n_all_vars,
+		 double *means)
 {
   size_t i;
   size_t j;
-  double weight;
-  double mean;
-  double variance;
-  double skewness;
-  double kurtosis;
-  /*
-     Scan the variable names in the columns of the design matrix.
-     When we find the variable we need, insert its mean in the cache.
-   */
-  for (i = 0; i < dm->m->size2; i++)
+  size_t dep_subscript;
+  size_t *rows;
+  const gsl_matrix *ssizes;
+  const gsl_matrix *cm;
+  const gsl_matrix *mean_matrix;
+  double result = 0.0;
+  
+  cm = covariance_calculate (all_cov);
+  rows = xnmalloc (cov->size1 - 1, sizeof (*rows));
+  
+  for (i = 0; i < n_all_vars; i++)
     {
-      for (j = 0; j < n; j++)
+      for (j = 0; j < n_vars; j++)
 	{
-	  if (design_matrix_col_to_var (dm, i) == (mom + j)->v)
+	  if (vars[j] == all_vars[i])
 	    {
-	      moments1_calculate ((mom + j)->m, &weight, &mean, &variance,
-				  &skewness, &kurtosis);
-	      gsl_vector_set (c->indep_means, i, mean);
-	      gsl_vector_set (c->indep_std, i, sqrt (variance));
+	      rows[j] = i;
 	    }
 	}
+      if (all_vars[i] == dep_var)
+	{
+	  dep_subscript = i;
+	}
+    }
+  mean_matrix = covariance_moments (all_cov, MOMENT_MEAN);
+  for (i = 0; i < cov->size1 - 1; i++)
+    {
+      means[i] = gsl_matrix_get (mean_matrix, rows[i], 0);
+      for (j = 0; j < cov->size2 - 1; j++)
+	{
+	  gsl_matrix_set (cov, i, j, gsl_matrix_get (cm, rows[i], rows[j]));
+	  gsl_matrix_set (cov, j, i, gsl_matrix_get (cm, rows[j], rows[i]));
+	}
     }
+  means[cov->size1 - 1] = gsl_matrix_get (mean_matrix, dep_subscript, 0);
+  ssizes = covariance_moments (all_cov, MOMENT_NONE);
+  result = gsl_matrix_get (ssizes, dep_subscript, rows[0]);
+  for (i = 0; i < cov->size1 - 1; i++)
+    {
+      gsl_matrix_set (cov, i, cov->size1 - 1, 
+		      gsl_matrix_get (cm, rows[i], dep_subscript));
+      gsl_matrix_set (cov, cov->size1 - 1, i, 
+		      gsl_matrix_get (cm, rows[i], dep_subscript));
+      if (result > gsl_matrix_get (ssizes, rows[i], dep_subscript))
+	{
+	  result = gsl_matrix_get (ssizes, rows[i], dep_subscript);
+	}
+    }
+  gsl_matrix_set (cov, cov->size1 - 1, cov->size1 - 1, 
+		  gsl_matrix_get (cm, dep_subscript, dep_subscript));
+  free (rows);
+  return result;
 }
+
 static bool
-run_regression (const struct ccase *first,
-		const struct casefile *cf, void *cmd_ UNUSED,
-		const struct dataset *ds)
+run_regression (struct casereader *input, struct cmd_regression *cmd,
+		struct dataset *ds, linreg **models)
 {
   size_t i;
-  size_t n_data = 0;		/* Number of valide cases. */
-  size_t n_cases;		/* Number of cases. */
-  size_t row;
-  size_t case_num;
   int n_indep = 0;
   int k;
-  /*
-     Keep track of the missing cases.
-   */
-  int *is_missing_case;
-  const union value *val;
-  struct casereader *r;
-  struct ccase c;
-  const struct variable **indep_vars;
-  struct design_matrix *X;
-  struct moments_var *mom;
-  gsl_vector *Y;
-
-  pspp_linreg_opts lopts;
+  double n_data;
+  double *means;
+  struct ccase *c;
+  struct covariance *cov;
+  const struct variable **vars;
+  const struct variable *dep_var;
+  struct casereader *reader;
+  const struct dictionary *dict;
+  gsl_matrix *this_cm;
 
   assert (models != NULL);
 
-  output_split_file_values (ds, first);
-
-  if (!v_variables)
-    {
-      dict_get_vars (dataset_dict (ds), &v_variables, &n_variables,
-		     1u << DC_SYSTEM);
-    }
-
-  n_cases = casefile_get_case_cnt (cf);
-
-  for (i = 0; i < cmd.n_dependent; i++)
+  for (i = 0; i < n_variables; i++)
     {
-      if (!var_is_numeric (cmd.v_dependent[i]))
+      if (!var_is_numeric (v_variables[i]))
 	{
-	  msg (SE, gettext ("Dependent variable must be numeric."));
-	  pspp_reg_rc = CMD_FAILURE;
-	  return true;
+	  msg (SE, _("REGRESSION requires numeric variables."));
+	  return false;
 	}
     }
 
-  is_missing_case = xnmalloc (n_cases, sizeof (*is_missing_case));
-  mom = xnmalloc (n_variables, sizeof (*mom));
-  for (i = 0; i < n_variables; i++)
+  c = casereader_peek (input, 0);
+  if (c == NULL)
     {
-      (mom + i)->m = moments1_create (MOMENT_VARIANCE);
-      (mom + i)->v = v_variables[i];
+      casereader_destroy (input);
+      return true;
     }
-  lopts.get_depvar_mean_std = 1;
+  output_split_file_values (ds, c);
+  case_unref (c);
 
-  for (k = 0; k < cmd.n_dependent; k++)
+  dict = dataset_dict (ds);
+  if (!v_variables)
     {
-      n_indep = get_n_indep ((const struct variable *) cmd.v_dependent[k]);
-      lopts.get_indep_mean_std = xnmalloc (n_indep, sizeof (int));
-      indep_vars = xnmalloc (n_indep, sizeof *indep_vars);
-      assert (indep_vars != NULL);
-
-      for (i = 0; i < n_cases; i++)
+      dict_get_vars (dict, &v_variables, &n_variables, 0);
+    }
+  vars = xnmalloc (n_variables, sizeof (*vars));
+  means  = xnmalloc (n_variables, sizeof (*means));
+  cov = covariance_1pass_create (n_variables, v_variables,
+				 dict_get_weight (dict), MV_ANY);
+
+  reader = casereader_clone (input);
+  reader = casereader_create_filter_missing (reader, v_variables, n_variables,
+					     MV_ANY, NULL, NULL);
+  for (; (c = casereader_read (reader)) != NULL; case_unref (c))
+    {
+      covariance_accumulate (cov, c);
+    }
+  
+  for (k = 0; k < cmd->n_dependent; k++)
+    {
+      dep_var = cmd->v_dependent[k];
+      n_indep = identify_indep_vars (vars, dep_var);
+      
+      this_cm = gsl_matrix_alloc (n_indep + 1, n_indep + 1);
+      n_data = fill_covariance (this_cm, cov, vars, n_indep, 
+				dep_var, v_variables, n_variables, means);
+      models[k] = linreg_alloc (dep_var, (const struct variable **) vars,
+				n_data, n_indep);
+      models[k]->depvar = dep_var;
+      for (i = 0; i < n_indep; i++)
 	{
-	  is_missing_case[i] = 0;
+	  linreg_set_indep_variable_mean (models[k], i, means[i]);
 	}
-      n_data = prepare_data (n_cases, is_missing_case, indep_vars,
-			     cmd.v_dependent[k],
-			     (const struct casefile *) cf, mom);
+      /*
+	For large data sets, use QR decomposition.
+      */
+      if (n_data > sqrt (n_indep) && n_data > REG_LARGE_DATA)
+	{
+	  models[k]->method = LINREG_QR;
+	}
+      
       if (n_data > 0)
 	{
-	  Y = gsl_vector_alloc (n_data);
-	  X =
-	    design_matrix_create (n_indep, (const struct variable **) indep_vars,
-				  n_data);
-	  for (i = 0; i < X->m->size2; i++)
-	    {
-	      lopts.get_indep_mean_std[i] = 1;
-	    }
-	  models[k] = pspp_linreg_cache_alloc (X->m->size1, X->m->size2);
-	  models[k]->indep_means = gsl_vector_alloc (X->m->size2);
-	  models[k]->indep_std = gsl_vector_alloc (X->m->size2);
-	  models[k]->depvar = (const struct variable *) cmd.v_dependent[k];
 	  /*
-	    For large data sets, use QR decomposition.
+	    Find the least-squares estimates and other statistics.
 	  */
-	  if (n_data > sqrt (n_indep) && n_data > REG_LARGE_DATA)
-	    {
-	      models[k]->method = PSPP_LINREG_SVD;
-	    }
+	  linreg_fit (this_cm, models[k]);
 	  
-	  /*
-	    The second pass fills the design matrix.
-	  */
-	  row = 0;
-	  for (r = casefile_get_reader (cf, NULL); casereader_read (r, &c);
-	       case_destroy (&c))
-	    /* Iterate over the cases. */
+	  if (!taint_has_tainted_successor (casereader_get_taint (input)))
 	    {
-	      case_num = casereader_cnum (r) - 1;
-	      if (!is_missing_case[case_num])
-		{
-		  for (i = 0; i < n_variables; ++i)	/* Iterate over the
-							   variables for the
-							   current case.
-							*/
-		    {
-		      val = case_data (&c, v_variables[i]);
-		      /*
-			Independent/dependent variable separation. The
-			'variables' subcommand specifies a varlist which contains
-			both dependent and independent variables. The dependent
-			variables are specified with the 'dependent'
-			subcommand, and maybe also in the 'variables' subcommand. 
-			We need to separate the two.
-		      */
-		      if (!is_depvar (i, cmd.v_dependent[k]))
-			{
-			  if (var_is_alpha (v_variables[i]))
-			    {
-			      design_matrix_set_categorical (X, row,
-							     v_variables[i], val);
-			    }
-			  else
-			    {
-			      design_matrix_set_numeric (X, row, v_variables[i],
-							 val);
-			    }
-			}
-		    }
-		  val = case_data (&c, cmd.v_dependent[k]);
-		  gsl_vector_set (Y, row, val->f);
-		  row++;
-		}
+	      subcommand_statistics (cmd->a_statistics, models[k]);
 	    }
-	  /*
-	    Now that we know the number of coefficients, allocate space
-	    and store pointers to the variables that correspond to the
-	    coefficients.
-	  */
-	  coeff_init (models[k], X);
-	  
-	  /* 
-	     Find the least-squares estimates and other statistics.
-	  */
-	  pspp_linreg ((const gsl_vector *) Y, X->m, &lopts, models[k]);
-	  compute_moments (models[k], mom, X, n_variables);
-	  subcommand_statistics (cmd.a_statistics, models[k]);
-	  subcommand_export (cmd.sbc_export, models[k]);
-	  
-	  gsl_vector_free (Y);
-	  design_matrix_destroy (X);
-	  free (indep_vars);
-	  free (lopts.get_indep_mean_std);
-	  casereader_destroy (r);
+	}
+      else
+	{
+	  msg (SE,
+	       gettext ("No valid data found. This command was skipped."));
+	  linreg_free (models[k]);
+	  models[k] = NULL;
 	}
     }
-  for (i = 0; i < n_variables; i++)
-    {
-      moments1_destroy ((mom + i)->m);
-    }
-  free (mom);
-  free (is_missing_case);
-
+  
+  casereader_destroy (reader);
+  free (vars);
+  free (means);
+  casereader_destroy (input);
+  covariance_destroy (cov);
+  
   return true;
 }
 
 /*
-  Local Variables:   
+  Local Variables:
   mode: c
   End:
 */