1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2009 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 <math/covariance.h>
20 #include <math/design-matrix.h>
21 #include <gsl/gsl_matrix.h>
22 #include <data/casegrouper.h>
23 #include <data/casereader.h>
24 #include <data/dictionary.h>
25 #include <data/procedure.h>
26 #include <data/variable.h>
27 #include <language/command.h>
28 #include <language/dictionary/split-file.h>
29 #include <language/lexer/lexer.h>
30 #include <language/lexer/variable-parser.h>
31 #include <output/manager.h>
32 #include <output/table.h>
33 #include <libpspp/message.h>
34 #include <data/format.h>
35 #include <math/moments.h>
40 #include <libpspp/misc.h>
41 #include <gsl/gsl_cdf.h>
44 #define _(msgid) gettext (msgid)
45 #define N_(msgid) msgid
49 significance_of_correlation (double rho, double w)
52 t /= 1 - MIN (1, pow2 (rho));
57 return gsl_cdf_tdist_Q (t, w - 2);
59 return gsl_cdf_tdist_P (t, w - 2);
68 const struct variable **vars;
72 /* Handling of missing values. */
73 enum corr_missing_type
75 CORR_PAIRWISE, /* Handle missing values on a per-variable-pair basis. */
76 CORR_LISTWISE /* Discard entire case if any variable is missing. */
81 enum corr_missing_type missing_type;
82 enum mv_class exclude; /* Classes of missing values to exclude. */
84 bool sig; /* Flag significant values or not */
85 int tails; /* Report significance with how many tails ? */
87 const struct variable *wv; /* The weight variable (if any) */
92 output_correlation (const struct corr *corr, const struct corr_opts *opts,
93 const gsl_matrix *cm, const gsl_matrix *samples)
98 int nr = corr->n_vars1;
99 int nc = matrix_cols = corr->n_vars_total > corr->n_vars1 ?
100 corr->n_vars_total - corr->n_vars1 : corr->n_vars1;
102 const struct fmt_spec *wfmt = opts->wv ? var_get_print_format (opts->wv) : & F_8_0;
104 const int heading_columns = 2;
105 const int heading_rows = 1;
107 const int rows_per_variable = opts->missing_type == CORR_LISTWISE ? 2 : 3;
109 /* Two header columns */
110 nc += heading_columns;
112 /* Three data per variable */
113 nr *= rows_per_variable;
118 t = tab_create (nc, nr, 0);
119 tab_title (t, _("Correlations"));
120 tab_dim (t, tab_natural_dimensions, NULL);
122 tab_headers (t, heading_columns, 0, heading_rows, 0);
124 /* Outline the box */
138 tab_vline (t, TAL_2, heading_columns, 0, nr - 1);
139 tab_vline (t, TAL_1, 1, heading_rows, nr - 1);
141 for (r = 0 ; r < corr->n_vars1 ; ++r)
143 tab_text (t, 0, 1 + r * rows_per_variable, TAB_LEFT | TAT_TITLE,
144 var_to_string (corr->vars[r]));
146 tab_text (t, 1, 1 + r * rows_per_variable, TAB_LEFT | TAT_TITLE, _("Pearson Correlation"));
147 tab_text (t, 1, 2 + r * rows_per_variable, TAB_LEFT | TAT_TITLE,
148 (opts->tails == 2) ? _("Sig. (2-tailed)") : _("Sig. (1-tailed)"));
149 if ( opts->missing_type != CORR_LISTWISE )
150 tab_text (t, 1, 3 + r * rows_per_variable, TAB_LEFT | TAT_TITLE, _("N"));
151 tab_hline (t, TAL_1, 0, nc - 1, r * rows_per_variable + 1);
154 for (c = 0 ; c < matrix_cols ; ++c)
156 const struct variable *v = corr->n_vars_total > corr->n_vars1 ? corr->vars[corr->n_vars_total - corr->n_vars1 + c] : corr->vars[c];
157 tab_text (t, heading_columns + c, 0, TAB_LEFT | TAT_TITLE, var_to_string (v));
160 for (r = 0 ; r < corr->n_vars1 ; ++r)
162 const int row = r * rows_per_variable + heading_rows;
163 for (c = 0 ; c < matrix_cols ; ++c)
165 unsigned char flags = 0;
166 int col_index = corr->n_vars_total - corr->n_vars1 + c;
167 double pearson = gsl_matrix_get (cm, r, col_index);
168 double w = gsl_matrix_get (samples, r, col_index);
169 double sig = opts->tails * significance_of_correlation (pearson, w);
171 if ( opts->missing_type != CORR_LISTWISE )
172 tab_double (t, c + heading_columns, row + 2, 0, w, wfmt);
175 tab_double (t, c + heading_columns, row + 1, 0, sig, NULL);
177 if ( opts->sig && c != r && sig < 0.05)
180 tab_double (t, c + heading_columns, row, flags, pearson, NULL);
189 correlation_from_covariance (const gsl_matrix *cv, const gsl_matrix *v)
192 gsl_matrix *corr = gsl_matrix_calloc (cv->size1, cv->size2);
194 for (i = 0 ; i < cv->size1; ++i)
196 for (j = 0 ; j < cv->size2; ++j)
198 double rho = gsl_matrix_get (cv, i, j);
200 rho /= sqrt (gsl_matrix_get (v, i, j))
202 sqrt (gsl_matrix_get (v, j, i));
204 gsl_matrix_set (corr, i, j, rho);
215 run_corr (struct casereader *r, const struct corr_opts *opts, const struct corr *corr)
218 const gsl_matrix *var_matrix;
219 const gsl_matrix *samples_matrix;
220 const gsl_matrix *cov_matrix;
221 gsl_matrix *corr_matrix;
222 struct covariance *cov = covariance_create (corr->n_vars_total, corr->vars,
223 opts->wv, opts->exclude);
225 for ( ; (c = casereader_read (r) ); case_unref (c))
227 covariance_accumulate (cov, c);
230 cov_matrix = covariance_calculate (cov);
232 samples_matrix = covariance_moments (cov, MOMENT_NONE);
233 var_matrix = covariance_moments (cov, MOMENT_VARIANCE);
235 corr_matrix = correlation_from_covariance (cov_matrix, var_matrix);
237 output_correlation (corr, opts,
241 covariance_destroy (cov);
242 gsl_matrix_free (corr_matrix);
246 cmd_correlation (struct lexer *lexer, struct dataset *ds)
249 int n_all_vars = 0; /* Total number of variables involved in this command */
250 const struct variable **all_vars ;
251 const struct dictionary *dict = dataset_dict (ds);
254 struct casegrouper *grouper;
255 struct casereader *group;
257 struct corr *corr = NULL;
260 struct corr_opts opts;
261 opts.missing_type = CORR_PAIRWISE;
262 opts.wv = dict_get_weight (dict);
265 opts.exclude = MV_ANY;
267 /* Parse CORRELATIONS. */
268 while (lex_token (lexer) != '.')
270 lex_match (lexer, '/');
271 if (lex_match_id (lexer, "MISSING"))
273 lex_match (lexer, '=');
274 while (lex_token (lexer) != '.' && lex_token (lexer) != '/')
276 if (lex_match_id (lexer, "PAIRWISE"))
277 opts.missing_type = CORR_PAIRWISE;
278 else if (lex_match_id (lexer, "LISTWISE"))
279 opts.missing_type = CORR_LISTWISE;
281 else if (lex_match_id (lexer, "INCLUDE"))
282 opts.exclude = MV_SYSTEM;
283 else if (lex_match_id (lexer, "EXCLUDE"))
284 opts.exclude = MV_ANY;
287 lex_error (lexer, NULL);
290 lex_match (lexer, ',');
293 else if (lex_match_id (lexer, "PRINT"))
295 lex_match (lexer, '=');
296 while (lex_token (lexer) != '.' && lex_token (lexer) != '/')
298 if ( lex_match_id (lexer, "TWOTAIL"))
300 else if (lex_match_id (lexer, "ONETAIL"))
302 else if (lex_match_id (lexer, "SIG"))
304 else if (lex_match_id (lexer, "NOSIG"))
308 lex_error (lexer, NULL);
312 lex_match (lexer, ',');
317 if (lex_match_id (lexer, "VARIABLES"))
319 lex_match (lexer, '=');
322 corr = xrealloc (corr, sizeof (*corr) * (n_corrs + 1));
323 corr[n_corrs].n_vars_total = corr[n_corrs].n_vars1 = 0;
325 if ( ! parse_variables_const (lexer, dict, &corr[n_corrs].vars,
326 &corr[n_corrs].n_vars_total,
334 corr[n_corrs].n_vars1 = corr[n_corrs].n_vars_total;
336 if ( lex_match (lexer, T_WITH))
338 if ( ! parse_variables_const (lexer, dict,
339 &corr[n_corrs].vars, &corr[n_corrs].n_vars_total,
340 PV_NUMERIC | PV_APPEND))
347 n_all_vars += corr[n_corrs].n_vars_total;
355 msg (SE, _("No variables specified."));
360 all_vars = xmalloc (sizeof (*all_vars) * n_all_vars);
363 /* FIXME: Using a hash here would make more sense */
364 const struct variable **vv = all_vars;
366 for (i = 0 ; i < n_corrs; ++i)
369 const struct corr *c = &corr[i];
370 for (v = 0 ; v < c->n_vars_total; ++v)
375 grouper = casegrouper_create_splits (proc_open (ds), dict);
377 while (casegrouper_get_next_group (grouper, &group))
379 for (i = 0 ; i < n_corrs; ++i)
381 /* FIXME: No need to iterate the data multiple times */
382 struct casereader *r = casereader_clone (group);
384 if ( opts.missing_type == CORR_LISTWISE)
385 r = casereader_create_filter_missing (r, all_vars, n_all_vars,
386 opts.exclude, NULL, NULL);
389 run_corr (r, &opts, &corr[i]);
390 casereader_destroy (r);
392 casereader_destroy (group);
395 ok = casegrouper_destroy (grouper);
396 ok = proc_commit (ds) && ok;
403 return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;