1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 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 "language/stats/sign.h"
21 #include <gsl/gsl_cdf.h>
22 #include <gsl/gsl_randist.h>
24 #include "data/casereader.h"
25 #include "data/dataset.h"
26 #include "data/dictionary.h"
27 #include "data/format.h"
28 #include "data/missing-values.h"
29 #include "data/variable.h"
30 #include "language/stats/npar.h"
31 #include "libpspp/str.h"
32 #include "output/tab.h"
34 #include "gl/minmax.h"
35 #include "gl/xalloc.h"
38 #define _(msgid) gettext (msgid)
40 struct sign_test_params
46 double one_tailed_sig;
52 output_frequency_table (const struct two_sample_test *t2s,
53 const struct sign_test_params *param,
54 const struct dictionary *dict)
57 struct tab_table *table = tab_create (3, 1 + 4 * t2s->n_pairs);
59 const struct variable *wv = dict_get_weight (dict);
60 const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : & F_8_0;
62 tab_set_format (table, RC_WEIGHT, wfmt);
63 tab_title (table, _("Frequencies"));
65 tab_headers (table, 2, 0, 1, 0);
67 /* Vertical lines inside the box */
68 tab_box (table, 0, 0, -1, TAL_1,
69 1, 0, tab_nc (table) - 1, tab_nr (table) - 1 );
71 /* Box around entire table */
72 tab_box (table, TAL_2, TAL_2, -1, -1,
73 0, 0, tab_nc (table) - 1, tab_nr (table) - 1 );
75 tab_text (table, 2, 0, TAB_CENTER, _("N"));
77 for (i = 0 ; i < t2s->n_pairs; ++i)
79 variable_pair *vp = &t2s->pairs[i];
81 struct string pair_name;
82 ds_init_cstr (&pair_name, var_to_string ((*vp)[0]));
83 ds_put_cstr (&pair_name, " - ");
84 ds_put_cstr (&pair_name, var_to_string ((*vp)[1]));
87 tab_text (table, 0, 1 + i * 4, TAB_LEFT, ds_cstr (&pair_name));
89 ds_destroy (&pair_name);
91 tab_hline (table, TAL_1, 0, tab_nc (table) - 1, 1 + i * 4);
93 tab_text (table, 1, 1 + i * 4, TAB_LEFT, _("Negative Differences"));
94 tab_text (table, 1, 2 + i * 4, TAB_LEFT, _("Positive Differences"));
95 tab_text (table, 1, 3 + i * 4, TAB_LEFT, _("Ties"));
96 tab_text (table, 1, 4 + i * 4, TAB_LEFT, _("Total"));
98 tab_double (table, 2, 1 + i * 4, TAB_RIGHT, param[i].neg, NULL, RC_WEIGHT);
99 tab_double (table, 2, 2 + i * 4, TAB_RIGHT, param[i].pos, NULL, RC_WEIGHT);
100 tab_double (table, 2, 3 + i * 4, TAB_RIGHT, param[i].ties, NULL, RC_WEIGHT);
101 tab_double (table, 2, 4 + i * 4, TAB_RIGHT,
102 param[i].ties + param[i].neg + param[i].pos, NULL, RC_WEIGHT);
109 output_statistics_table (const struct two_sample_test *t2s,
110 const struct sign_test_params *param)
113 struct tab_table *table = tab_create (1 + t2s->n_pairs, 4);
115 tab_title (table, _("Test Statistics"));
117 tab_headers (table, 0, 1, 0, 1);
119 tab_hline (table, TAL_2, 0, tab_nc (table) - 1, 1);
120 tab_vline (table, TAL_2, 1, 0, tab_nr (table) - 1);
123 /* Vertical lines inside the box */
124 tab_box (table, -1, -1, -1, TAL_1,
126 tab_nc (table) - 1, tab_nr (table) - 1);
128 /* Box around entire table */
129 tab_box (table, TAL_2, TAL_2, -1, -1,
130 0, 0, tab_nc (table) - 1,
133 tab_text (table, 0, 1, TAT_TITLE | TAB_LEFT,
134 _("Exact Sig. (2-tailed)"));
136 tab_text (table, 0, 2, TAT_TITLE | TAB_LEFT,
137 _("Exact Sig. (1-tailed)"));
139 tab_text (table, 0, 3, TAT_TITLE | TAB_LEFT,
140 _("Point Probability"));
142 for (i = 0 ; i < t2s->n_pairs; ++i)
144 variable_pair *vp = &t2s->pairs[i];
146 struct string pair_name;
147 ds_init_cstr (&pair_name, var_to_string ((*vp)[0]));
148 ds_put_cstr (&pair_name, " - ");
149 ds_put_cstr (&pair_name, var_to_string ((*vp)[1]));
151 tab_text (table, 1 + i, 0, TAB_LEFT, ds_cstr (&pair_name));
152 ds_destroy (&pair_name);
154 tab_double (table, 1 + i, 1, TAB_RIGHT,
155 param[i].one_tailed_sig * 2, NULL, RC_PVALUE);
157 tab_double (table, 1 + i, 2, TAB_RIGHT, param[i].one_tailed_sig, NULL, RC_PVALUE);
158 tab_double (table, 1 + i, 3, TAB_RIGHT, param[i].point_prob, NULL, RC_PVALUE);
165 sign_execute (const struct dataset *ds,
166 struct casereader *input,
167 enum mv_class exclude,
168 const struct npar_test *test,
174 const struct dictionary *dict = dataset_dict (ds);
175 const struct two_sample_test *t2s = UP_CAST (test, const struct two_sample_test, parent);
178 struct sign_test_params *stp = xcalloc (t2s->n_pairs, sizeof *stp);
180 struct casereader *r = input;
182 for (; (c = casereader_read (r)) != NULL; case_unref (c))
184 const double weight = dict_get_case_weight (dict, c, &warn);
186 for (i = 0 ; i < t2s->n_pairs; ++i )
188 variable_pair *vp = &t2s->pairs[i];
189 const union value *value0 = case_data (c, (*vp)[0]);
190 const union value *value1 = case_data (c, (*vp)[1]);
191 const double diff = value0->f - value1->f;
193 if (var_is_value_missing ((*vp)[0], value0, exclude))
196 if (var_is_value_missing ((*vp)[1], value1, exclude))
200 stp[i].pos += weight;
202 stp[i].neg += weight;
204 stp[i].ties += weight;
208 casereader_destroy (r);
210 for (i = 0 ; i < t2s->n_pairs; ++i )
212 int r = MIN (stp[i].pos, stp[i].neg);
213 stp[i].one_tailed_sig = gsl_cdf_binomial_P (r,
215 stp[i].pos + stp[i].neg);
217 stp[i].point_prob = gsl_ran_binomial_pdf (r, 0.5,
218 stp[i].pos + stp[i].neg);
221 output_frequency_table (t2s, stp, dict);
223 output_statistics_table (t2s, stp);