1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2005, 2009, 2010, 2011 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19 #include <gsl/gsl_cdf.h>
20 #include <gsl/gsl_matrix.h>
21 #include <gsl/gsl_vector.h>
25 #include "data/case.h"
26 #include "data/casegrouper.h"
27 #include "data/casereader.h"
28 #include "data/dataset.h"
29 #include "data/dictionary.h"
30 #include "data/missing-values.h"
31 #include "data/transformations.h"
32 #include "data/value-labels.h"
33 #include "data/variable.h"
34 #include "language/command.h"
35 #include "language/data-io/file-handle.h"
36 #include "language/dictionary/split-file.h"
37 #include "language/lexer/lexer.h"
38 #include "libpspp/compiler.h"
39 #include "libpspp/message.h"
40 #include "libpspp/taint.h"
41 #include "math/covariance.h"
42 #include "math/linreg.h"
43 #include "math/moments.h"
44 #include "output/tab.h"
46 #include "gl/intprops.h"
47 #include "gl/xalloc.h"
50 #define _(msgid) gettext (msgid)
52 #define REG_LARGE_DATA 1000
57 "REGRESSION" (regression_):
77 +save[sv_]=resid,pred;
82 static struct cmd_regression cmd;
85 Moments for each of the variables used.
90 const struct variable *v;
94 Transformations for saving predicted values
99 int n_trns; /* Number of transformations. */
100 int trns_id; /* Which trns is this one? */
101 linreg *c; /* Linear model for this trns. */
104 Variables used (both explanatory and response).
106 static const struct variable **v_variables;
111 static size_t n_variables;
113 static bool run_regression (struct casereader *, struct cmd_regression *,
114 struct dataset *, linreg **);
117 STATISTICS subcommand output functions.
119 static void reg_stats_r (linreg *, void *);
120 static void reg_stats_coeff (linreg *, void *);
121 static void reg_stats_anova (linreg *, void *);
122 static void reg_stats_outs (linreg *, void *);
123 static void reg_stats_zpp (linreg *, void *);
124 static void reg_stats_label (linreg *, void *);
125 static void reg_stats_sha (linreg *, void *);
126 static void reg_stats_ci (linreg *, void *);
127 static void reg_stats_f (linreg *, void *);
128 static void reg_stats_bcov (linreg *, void *);
129 static void reg_stats_ses (linreg *, void *);
130 static void reg_stats_xtx (linreg *, void *);
131 static void reg_stats_collin (linreg *, void *);
132 static void reg_stats_tol (linreg *, void *);
133 static void reg_stats_selection (linreg *, void *);
134 static void statistics_keyword_output (void (*)(linreg *, void *),
135 int, linreg *, void *);
138 reg_stats_r (linreg *c, void *aux UNUSED)
148 rsq = linreg_ssreg (c) / linreg_sst (c);
149 adjrsq = 1.0 - (1.0 - rsq) * (linreg_n_obs (c) - 1.0) / (linreg_n_obs (c) - linreg_n_coeffs (c));
150 std_error = sqrt (linreg_mse (c));
151 t = tab_create (n_cols, n_rows);
152 tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
153 tab_hline (t, TAL_2, 0, n_cols - 1, 1);
154 tab_vline (t, TAL_2, 2, 0, n_rows - 1);
155 tab_vline (t, TAL_0, 1, 0, 0);
157 tab_text (t, 1, 0, TAB_CENTER | TAT_TITLE, _("R"));
158 tab_text (t, 2, 0, TAB_CENTER | TAT_TITLE, _("R Square"));
159 tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("Adjusted R Square"));
160 tab_text (t, 4, 0, TAB_CENTER | TAT_TITLE, _("Std. Error of the Estimate"));
161 tab_double (t, 1, 1, TAB_RIGHT, sqrt (rsq), NULL);
162 tab_double (t, 2, 1, TAB_RIGHT, rsq, NULL);
163 tab_double (t, 3, 1, TAB_RIGHT, adjrsq, NULL);
164 tab_double (t, 4, 1, TAB_RIGHT, std_error, NULL);
165 tab_title (t, _("Model Summary"));
170 Table showing estimated regression coefficients.
173 reg_stats_coeff (linreg * c, void *aux_)
185 const struct variable *v;
187 gsl_matrix *cov = aux_;
190 n_rows = linreg_n_coeffs (c) + 3;
192 t = tab_create (n_cols, n_rows);
193 tab_headers (t, 2, 0, 1, 0);
194 tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
195 tab_hline (t, TAL_2, 0, n_cols - 1, 1);
196 tab_vline (t, TAL_2, 2, 0, n_rows - 1);
197 tab_vline (t, TAL_0, 1, 0, 0);
199 tab_text (t, 2, 0, TAB_CENTER | TAT_TITLE, _("B"));
200 tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("Std. Error"));
201 tab_text (t, 4, 0, TAB_CENTER | TAT_TITLE, _("Beta"));
202 tab_text (t, 5, 0, TAB_CENTER | TAT_TITLE, _("t"));
203 tab_text (t, 6, 0, TAB_CENTER | TAT_TITLE, _("Significance"));
204 tab_text (t, 1, 1, TAB_LEFT | TAT_TITLE, _("(Constant)"));
205 tab_double (t, 2, 1, 0, linreg_intercept (c), NULL);
206 std_err = sqrt (gsl_matrix_get (linreg_cov (c), 0, 0));
207 tab_double (t, 3, 1, 0, std_err, NULL);
208 tab_double (t, 4, 1, 0, 0.0, NULL);
209 t_stat = linreg_intercept (c) / std_err;
210 tab_double (t, 5, 1, 0, t_stat, NULL);
211 pval = 2 * gsl_cdf_tdist_Q (fabs (t_stat), (double) (linreg_n_obs (c) - linreg_n_coeffs (c)));
212 tab_double (t, 6, 1, 0, pval, NULL);
213 for (j = 0; j < linreg_n_coeffs (c); j++)
216 ds_init_empty (&tstr);
219 v = linreg_indep_var (c, j);
220 label = var_to_string (v);
221 /* Do not overwrite the variable's name. */
222 ds_put_cstr (&tstr, label);
223 tab_text (t, 1, this_row, TAB_CENTER, ds_cstr (&tstr));
225 Regression coefficients.
227 tab_double (t, 2, this_row, 0, linreg_coeff (c, j), NULL);
229 Standard error of the coefficients.
231 std_err = sqrt (gsl_matrix_get (linreg_cov (c), j + 1, j + 1));
232 tab_double (t, 3, this_row, 0, std_err, NULL);
234 Standardized coefficient, i.e., regression coefficient
235 if all variables had unit variance.
237 beta = sqrt (gsl_matrix_get (cov, j, j));
238 beta *= linreg_coeff (c, j) /
239 sqrt (gsl_matrix_get (cov, cov->size1 - 1, cov->size2 - 1));
240 tab_double (t, 4, this_row, 0, beta, NULL);
243 Test statistic for H0: coefficient is 0.
245 t_stat = linreg_coeff (c, j) / std_err;
246 tab_double (t, 5, this_row, 0, t_stat, NULL);
248 P values for the test statistic above.
251 2 * gsl_cdf_tdist_Q (fabs (t_stat),
252 (double) (linreg_n_obs (c) - linreg_n_coeffs (c)));
253 tab_double (t, 6, this_row, 0, pval, NULL);
256 tab_title (t, _("Coefficients"));
261 Display the ANOVA table.
264 reg_stats_anova (linreg * c, void *aux UNUSED)
268 const double msm = linreg_ssreg (c) / linreg_dfmodel (c);
269 const double mse = linreg_mse (c);
270 const double F = msm / mse;
271 const double pval = gsl_cdf_fdist_Q (F, c->dfm, c->dfe);
276 t = tab_create (n_cols, n_rows);
277 tab_headers (t, 2, 0, 1, 0);
279 tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
281 tab_hline (t, TAL_2, 0, n_cols - 1, 1);
282 tab_vline (t, TAL_2, 2, 0, n_rows - 1);
283 tab_vline (t, TAL_0, 1, 0, 0);
285 tab_text (t, 2, 0, TAB_CENTER | TAT_TITLE, _("Sum of Squares"));
286 tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("df"));
287 tab_text (t, 4, 0, TAB_CENTER | TAT_TITLE, _("Mean Square"));
288 tab_text (t, 5, 0, TAB_CENTER | TAT_TITLE, _("F"));
289 tab_text (t, 6, 0, TAB_CENTER | TAT_TITLE, _("Significance"));
291 tab_text (t, 1, 1, TAB_LEFT | TAT_TITLE, _("Regression"));
292 tab_text (t, 1, 2, TAB_LEFT | TAT_TITLE, _("Residual"));
293 tab_text (t, 1, 3, TAB_LEFT | TAT_TITLE, _("Total"));
295 /* Sums of Squares */
296 tab_double (t, 2, 1, 0, linreg_ssreg (c), NULL);
297 tab_double (t, 2, 3, 0, linreg_sst (c), NULL);
298 tab_double (t, 2, 2, 0, linreg_sse (c), NULL);
301 /* Degrees of freedom */
302 tab_text_format (t, 3, 1, TAB_RIGHT, "%g", c->dfm);
303 tab_text_format (t, 3, 2, TAB_RIGHT, "%g", c->dfe);
304 tab_text_format (t, 3, 3, TAB_RIGHT, "%g", c->dft);
307 tab_double (t, 4, 1, TAB_RIGHT, msm, NULL);
308 tab_double (t, 4, 2, TAB_RIGHT, mse, NULL);
310 tab_double (t, 5, 1, 0, F, NULL);
312 tab_double (t, 6, 1, 0, pval, NULL);
314 tab_title (t, _("ANOVA"));
319 reg_stats_outs (linreg * c, void *aux UNUSED)
325 reg_stats_zpp (linreg * c, void *aux UNUSED)
331 reg_stats_label (linreg * c, void *aux UNUSED)
337 reg_stats_sha (linreg * c, void *aux UNUSED)
342 reg_stats_ci (linreg * c, void *aux UNUSED)
347 reg_stats_f (linreg * c, void *aux UNUSED)
352 reg_stats_bcov (linreg * c, void *aux UNUSED)
364 n_cols = c->n_indeps + 1 + 2;
365 n_rows = 2 * (c->n_indeps + 1);
366 t = tab_create (n_cols, n_rows);
367 tab_headers (t, 2, 0, 1, 0);
368 tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1);
369 tab_hline (t, TAL_2, 0, n_cols - 1, 1);
370 tab_vline (t, TAL_2, 2, 0, n_rows - 1);
371 tab_vline (t, TAL_0, 1, 0, 0);
372 tab_text (t, 0, 0, TAB_CENTER | TAT_TITLE, _("Model"));
373 tab_text (t, 1, 1, TAB_CENTER | TAT_TITLE, _("Covariances"));
374 for (i = 0; i < linreg_n_coeffs (c); i++)
376 const struct variable *v = linreg_indep_var (c, i);
377 label = var_to_string (v);
378 tab_text (t, 2, i, TAB_CENTER, label);
379 tab_text (t, i + 2, 0, TAB_CENTER, label);
380 for (k = 1; k < linreg_n_coeffs (c); k++)
382 col = (i <= k) ? k : i;
383 row = (i <= k) ? i : k;
384 tab_double (t, k + 2, i, TAB_CENTER,
385 gsl_matrix_get (c->cov, row, col), NULL);
388 tab_title (t, _("Coefficient Correlations"));
392 reg_stats_ses (linreg * c, void *aux UNUSED)
397 reg_stats_xtx (linreg * c, void *aux UNUSED)
402 reg_stats_collin (linreg * c, void *aux UNUSED)
407 reg_stats_tol (linreg * c, void *aux UNUSED)
412 reg_stats_selection (linreg * c, void *aux UNUSED)
418 statistics_keyword_output (void (*function) (linreg *, void *),
419 int keyword, linreg * c, void *aux)
423 (*function) (c, aux);
428 subcommand_statistics (int *keywords, linreg * c, void *aux)
431 The order here must match the order in which the STATISTICS
432 keywords appear in the specification section above.
459 Set everything but F.
461 for (i = 0; i < f; i++)
468 for (i = 0; i < all; i++)
476 Default output: ANOVA table, parameter estimates,
477 and statistics for variables not entered into model,
480 if (keywords[defaults] | d)
488 statistics_keyword_output (reg_stats_r, keywords[r], c, aux);
489 statistics_keyword_output (reg_stats_anova, keywords[anova], c, aux);
490 statistics_keyword_output (reg_stats_coeff, keywords[coeff], c, aux);
491 statistics_keyword_output (reg_stats_outs, keywords[outs], c, aux);
492 statistics_keyword_output (reg_stats_zpp, keywords[zpp], c, aux);
493 statistics_keyword_output (reg_stats_label, keywords[label], c, aux);
494 statistics_keyword_output (reg_stats_sha, keywords[sha], c, aux);
495 statistics_keyword_output (reg_stats_ci, keywords[ci], c, aux);
496 statistics_keyword_output (reg_stats_f, keywords[f], c, aux);
497 statistics_keyword_output (reg_stats_bcov, keywords[bcov], c, aux);
498 statistics_keyword_output (reg_stats_ses, keywords[ses], c, aux);
499 statistics_keyword_output (reg_stats_xtx, keywords[xtx], c, aux);
500 statistics_keyword_output (reg_stats_collin, keywords[collin], c, aux);
501 statistics_keyword_output (reg_stats_tol, keywords[tol], c, aux);
502 statistics_keyword_output (reg_stats_selection, keywords[selection], c, aux);
506 Free the transformation. Free its linear model if this
507 transformation is the last one.
510 regression_trns_free (void *t_)
513 struct reg_trns *t = t_;
515 if (t->trns_id == t->n_trns)
517 result = linreg_free (t->c);
525 Gets the predicted values.
528 regression_trns_pred_proc (void *t_, struct ccase **c,
529 casenumber case_idx UNUSED)
533 struct reg_trns *trns = t_;
535 union value *output = NULL;
536 const union value *tmp;
538 const struct variable **vars = NULL;
540 assert (trns != NULL);
542 assert (model != NULL);
543 assert (model->depvar != NULL);
544 assert (model->pred != NULL);
546 vars = linreg_get_vars (model);
547 n_vals = linreg_n_coeffs (model);
548 vals = xnmalloc (n_vals, sizeof (*vals));
549 *c = case_unshare (*c);
551 output = case_data_rw (*c, model->pred);
553 for (i = 0; i < n_vals; i++)
555 tmp = case_data (*c, vars[i]);
558 output->f = linreg_predict (model, vals, n_vals);
560 return TRNS_CONTINUE;
567 regression_trns_resid_proc (void *t_, struct ccase **c,
568 casenumber case_idx UNUSED)
572 struct reg_trns *trns = t_;
574 union value *output = NULL;
575 const union value *tmp;
578 const struct variable **vars = NULL;
580 assert (trns != NULL);
582 assert (model != NULL);
583 assert (model->depvar != NULL);
584 assert (model->resid != NULL);
586 vars = linreg_get_vars (model);
587 n_vals = linreg_n_coeffs (model);
589 vals = xnmalloc (n_vals, sizeof (*vals));
590 *c = case_unshare (*c);
591 output = case_data_rw (*c, model->resid);
592 assert (output != NULL);
594 for (i = 0; i < n_vals; i++)
596 tmp = case_data (*c, vars[i]);
599 tmp = case_data (*c, model->depvar);
601 output->f = linreg_residual (model, obs, vals, n_vals);
604 return TRNS_CONTINUE;
608 reg_get_name (const struct dictionary *dict, const char *prefix)
613 /* XXX handle too-long prefixes */
614 name = xmalloc (strlen (prefix) + INT_BUFSIZE_BOUND (i) + 1);
617 sprintf (name, "%s%d", prefix, i);
618 if (dict_lookup_var (dict, name) == NULL)
624 reg_save_var (struct dataset *ds, const char *prefix, trns_proc_func * f,
625 linreg * c, struct variable **v, int n_trns)
627 struct dictionary *dict = dataset_dict (ds);
628 static int trns_index = 1;
630 struct variable *new_var;
631 struct reg_trns *t = NULL;
633 t = xmalloc (sizeof (*t));
634 t->trns_id = trns_index;
638 name = reg_get_name (dict, prefix);
639 new_var = dict_create_var_assert (dict, name, 0);
643 add_transformation (ds, f, regression_trns_free, t);
647 subcommand_save (struct dataset *ds, int save, linreg ** models)
655 /* Count the number of transformations we will need. */
656 for (i = 0; i < REGRESSION_SV_count; i++)
663 n_trns *= cmd.n_dependent;
665 for (lc = models; lc < models + cmd.n_dependent; lc++)
669 if ((*lc)->depvar != NULL)
671 if (cmd.a_save[REGRESSION_SV_RESID])
673 reg_save_var (ds, "RES", regression_trns_resid_proc, *lc,
674 &(*lc)->resid, n_trns);
676 if (cmd.a_save[REGRESSION_SV_PRED])
678 reg_save_var (ds, "PRED", regression_trns_pred_proc, *lc,
679 &(*lc)->pred, n_trns);
687 for (lc = models; lc < models + cmd.n_dependent; lc++)
698 cmd_regression (struct lexer *lexer, struct dataset *ds)
700 struct casegrouper *grouper;
701 struct casereader *group;
706 if (!parse_regression (lexer, ds, &cmd, NULL))
711 models = xnmalloc (cmd.n_dependent, sizeof *models);
712 for (i = 0; i < cmd.n_dependent; i++)
718 grouper = casegrouper_create_splits (proc_open (ds), dataset_dict (ds));
719 while (casegrouper_get_next_group (grouper, &group))
720 run_regression (group, &cmd, ds, models);
721 ok = casegrouper_destroy (grouper);
722 ok = proc_commit (ds) && ok;
724 subcommand_save (ds, cmd.sbc_save, models);
727 free_regression (&cmd);
729 return ok ? CMD_SUCCESS : CMD_FAILURE;
733 Is variable k the dependent variable?
736 is_depvar (size_t k, const struct variable *v)
738 return v == v_variables[k];
741 /* Parser for the variables sub command */
743 regression_custom_variables (struct lexer *lexer, struct dataset *ds,
744 struct cmd_regression *cmd UNUSED,
747 const struct dictionary *dict = dataset_dict (ds);
749 lex_match (lexer, T_EQUALS);
751 if ((lex_token (lexer) != T_ID
752 || dict_lookup_var (dict, lex_tokcstr (lexer)) == NULL)
753 && lex_token (lexer) != T_ALL)
757 if (!parse_variables_const
758 (lexer, dict, &v_variables, &n_variables, PV_NONE))
763 assert (n_variables);
768 /* Identify the explanatory variables in v_variables. Returns
769 the number of independent variables. */
771 identify_indep_vars (const struct variable **indep_vars,
772 const struct variable *depvar)
774 int n_indep_vars = 0;
777 for (i = 0; i < n_variables; i++)
778 if (!is_depvar (i, depvar))
779 indep_vars[n_indep_vars++] = v_variables[i];
780 if ((n_indep_vars < 1) && is_depvar (0, depvar))
783 There is only one independent variable, and it is the same
784 as the dependent variable. Print a warning and continue.
787 gettext ("The dependent variable is equal to the independent variable."
788 "The least squares line is therefore Y=X."
789 "Standard errors and related statistics may be meaningless."));
791 indep_vars[0] = v_variables[0];
797 fill_covariance (gsl_matrix *cov, struct covariance *all_cov,
798 const struct variable **vars,
799 size_t n_vars, const struct variable *dep_var,
800 const struct variable **all_vars, size_t n_all_vars,
805 size_t dep_subscript;
807 const gsl_matrix *ssizes;
808 const gsl_matrix *mean_matrix;
809 const gsl_matrix *ssize_matrix;
812 gsl_matrix *cm = covariance_calculate_unnormalized (all_cov);
817 rows = xnmalloc (cov->size1 - 1, sizeof (*rows));
819 for (i = 0; i < n_all_vars; i++)
821 for (j = 0; j < n_vars; j++)
823 if (vars[j] == all_vars[i])
828 if (all_vars[i] == dep_var)
833 mean_matrix = covariance_moments (all_cov, MOMENT_MEAN);
834 ssize_matrix = covariance_moments (all_cov, MOMENT_NONE);
835 for (i = 0; i < cov->size1 - 1; i++)
837 means[i] = gsl_matrix_get (mean_matrix, rows[i], 0)
838 / gsl_matrix_get (ssize_matrix, rows[i], 0);
839 for (j = 0; j < cov->size2 - 1; j++)
841 gsl_matrix_set (cov, i, j, gsl_matrix_get (cm, rows[i], rows[j]));
842 gsl_matrix_set (cov, j, i, gsl_matrix_get (cm, rows[j], rows[i]));
845 means[cov->size1 - 1] = gsl_matrix_get (mean_matrix, dep_subscript, 0)
846 / gsl_matrix_get (ssize_matrix, dep_subscript, 0);
847 ssizes = covariance_moments (all_cov, MOMENT_NONE);
848 result = gsl_matrix_get (ssizes, dep_subscript, rows[0]);
849 for (i = 0; i < cov->size1 - 1; i++)
851 gsl_matrix_set (cov, i, cov->size1 - 1,
852 gsl_matrix_get (cm, rows[i], dep_subscript));
853 gsl_matrix_set (cov, cov->size1 - 1, i,
854 gsl_matrix_get (cm, rows[i], dep_subscript));
855 if (result > gsl_matrix_get (ssizes, rows[i], dep_subscript))
857 result = gsl_matrix_get (ssizes, rows[i], dep_subscript);
860 gsl_matrix_set (cov, cov->size1 - 1, cov->size1 - 1,
861 gsl_matrix_get (cm, dep_subscript, dep_subscript));
863 gsl_matrix_free (cm);
867 get_n_all_vars (struct cmd_regression *cmd)
869 size_t result = n_variables;
873 result += cmd->n_dependent;
874 for (i = 0; i < cmd->n_dependent; i++)
876 for (j = 0; j < n_variables; j++)
878 if (v_variables[j] == cmd->v_dependent[i])
887 fill_all_vars (const struct variable **vars, struct cmd_regression *cmd)
893 for (i = 0; i < n_variables; i++)
895 vars[i] = v_variables[i];
897 for (i = 0; i < cmd->n_dependent; i++)
900 for (j = 0; j < n_variables; j++)
902 if (cmd->v_dependent[i] == v_variables[j])
910 vars[i + n_variables] = cmd->v_dependent[i];
915 run_regression (struct casereader *input, struct cmd_regression *cmd,
916 struct dataset *ds, linreg **models)
924 struct covariance *cov;
925 const struct variable **vars;
926 const struct variable **all_vars;
927 const struct variable *dep_var;
928 struct casereader *reader;
929 const struct dictionary *dict;
932 assert (models != NULL);
934 for (i = 0; i < n_variables; i++)
936 if (!var_is_numeric (v_variables[i]))
938 msg (SE, _("REGRESSION requires numeric variables."));
943 c = casereader_peek (input, 0);
946 casereader_destroy (input);
949 output_split_file_values (ds, c);
952 dict = dataset_dict (ds);
955 dict_get_vars (dict, &v_variables, &n_variables, 0);
957 n_all_vars = get_n_all_vars (cmd);
958 all_vars = xnmalloc (n_all_vars, sizeof (*all_vars));
959 fill_all_vars (all_vars, cmd);
960 vars = xnmalloc (n_variables, sizeof (*vars));
961 means = xnmalloc (n_all_vars, sizeof (*means));
962 cov = covariance_1pass_create (n_all_vars, all_vars,
963 dict_get_weight (dict), MV_ANY);
965 reader = casereader_clone (input);
966 reader = casereader_create_filter_missing (reader, v_variables, n_variables,
968 for (; (c = casereader_read (reader)) != NULL; case_unref (c))
970 covariance_accumulate (cov, c);
973 for (k = 0; k < cmd->n_dependent; k++)
976 dep_var = cmd->v_dependent[k];
977 n_indep = identify_indep_vars (vars, dep_var);
979 this_cm = gsl_matrix_alloc (n_indep + 1, n_indep + 1);
980 n_data = fill_covariance (this_cm, cov, vars, n_indep,
981 dep_var, all_vars, n_all_vars, means);
982 models[k] = linreg_alloc (dep_var, (const struct variable **) vars,
984 models[k]->depvar = dep_var;
985 for (i = 0; i < n_indep; i++)
987 linreg_set_indep_variable_mean (models[k], i, means[i]);
989 linreg_set_depvar_mean (models[k], means[i]);
991 For large data sets, use QR decomposition.
993 if (n_data > sqrt (n_indep) && n_data > REG_LARGE_DATA)
995 models[k]->method = LINREG_QR;
1001 Find the least-squares estimates and other statistics.
1003 linreg_fit (this_cm, models[k]);
1005 if (!taint_has_tainted_successor (casereader_get_taint (input)))
1007 subcommand_statistics (cmd->a_statistics, models[k], this_cm);
1013 gettext ("No valid data found. This command was skipped."));
1014 linreg_free (models[k]);
1017 gsl_matrix_free (this_cm);
1020 casereader_destroy (reader);
1024 casereader_destroy (input);
1025 covariance_destroy (cov);