1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2006, 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/>. */
18 #include <libpspp/compiler.h>
19 #include <output/table.h>
21 #include <data/format.h>
22 #include <data/case.h>
23 #include <data/casereader.h>
24 #include <data/dictionary.h>
25 #include <data/procedure.h>
26 #include <data/variable.h>
27 #include <data/value.h>
28 #include <data/value-labels.h>
30 #include <libpspp/message.h>
31 #include <libpspp/assertion.h>
39 #define _(msgid) gettext (msgid)
41 #include <libpspp/misc.h>
43 #include <gsl/gsl_cdf.h>
44 #include <gsl/gsl_randist.h>
48 #include <libpspp/hash.h>
50 static double calculate_binomial_internal (double n1, double n2,
55 swap (double *i1, double *i2)
63 calculate_binomial (double n1, double n2, double p)
65 const double n = n1 + n2;
66 const bool test_reversed = (n1 / n > p ) ;
73 return calculate_binomial_internal (n1, n2, p);
77 calculate_binomial_internal (double n1, double n2, double p)
79 /* SPSS Statistical Algorithms has completely different and WRONG
82 double sig1tailed = gsl_cdf_binomial_P (n1, p, n1 + n2);
85 return sig1tailed > 0.5 ? 1.0 :sig1tailed * 2.0;
91 do_binomial (const struct dictionary *dict,
92 struct casereader *input,
93 const struct binomial_test *bst,
94 struct freq_mutable *cat1,
95 struct freq_mutable *cat2,
101 const struct one_sample_test *ost = (const struct one_sample_test *) bst;
104 while ((c = casereader_read(input)) != NULL)
107 double w = dict_get_case_weight (dict, c, &warn);
109 for (v = 0 ; v < ost->n_vars ; ++v )
111 const struct variable *var = ost->vars[v];
112 const union value *value = case_data (c, var);
113 int width = var_get_width (var);
115 if (var_is_value_missing (var, value, exclude))
118 if (bst->cutpoint != SYSMIS)
120 if ( compare_values_short (cat1[v].value, value, var) >= 0 )
127 if ( NULL == cat1[v].value )
129 cat1[v].value = value_dup (value, width);
132 else if ( 0 == compare_values_short (cat1[v].value, value, var))
134 else if ( NULL == cat2[v].value )
136 cat2[v].value = value_dup (value, width);
139 else if ( 0 == compare_values_short (cat2[v].value, value, var))
141 else if ( bst->category1 == SYSMIS)
142 msg (ME, _("Variable %s is not dichotomous"), var_get_name (var));
148 return casereader_destroy (input);
154 binomial_execute (const struct dataset *ds,
155 struct casereader *input,
156 enum mv_class exclude,
157 const struct npar_test *test,
162 const struct dictionary *dict = dataset_dict (ds);
163 const struct binomial_test *bst = (const struct binomial_test *) test;
164 const struct one_sample_test *ost = (const struct one_sample_test*) test;
166 struct freq_mutable *cat1 = xzalloc (sizeof (*cat1) * ost->n_vars);
167 struct freq_mutable *cat2 = xzalloc (sizeof (*cat1) * ost->n_vars);
169 assert ((bst->category1 == SYSMIS) == (bst->category2 == SYSMIS) || bst->cutpoint != SYSMIS);
171 if ( bst->cutpoint != SYSMIS )
176 for (i = 0; i < ost->n_vars; i++)
177 cat1[i].value = value_dup (&v, 0);
179 else if ( bst->category1 != SYSMIS )
183 v.f = bst->category1;
184 for (i = 0; i < ost->n_vars; i++)
185 cat1[i].value = value_dup (&v, 0);
188 if ( bst->category2 != SYSMIS )
192 v.f = bst->category2;
193 for (i = 0; i < ost->n_vars; i++)
194 cat2[i].value = value_dup (&v, 0);
197 if (do_binomial (dict, input, bst, cat1, cat2, exclude))
199 const struct variable *wvar = dict_get_weight (dict);
200 const struct fmt_spec *wfmt = wvar ?
201 var_get_print_format (wvar) : & F_8_0;
203 struct tab_table *table = tab_create (7, ost->n_vars * 3 + 1, 0);
205 tab_dim (table, tab_natural_dimensions);
207 tab_title (table, _("Binomial Test"));
209 tab_headers (table, 2, 0, 1, 0);
211 tab_box (table, TAL_1, TAL_1, -1, TAL_1,
212 0, 0, table->nc - 1, tab_nr(table) - 1 );
214 for (v = 0 ; v < ost->n_vars; ++v)
217 struct string catstr1;
218 struct string catstr2;
219 const struct variable *var = ost->vars[v];
221 ds_init_empty (&catstr1);
222 ds_init_empty (&catstr2);
224 if ( bst->cutpoint != SYSMIS)
226 ds_put_format (&catstr1, "<= %g", bst->cutpoint);
230 var_append_value_name (var, cat1[v].value, &catstr1);
231 var_append_value_name (var, cat2[v].value, &catstr2);
234 tab_hline (table, TAL_1, 0, tab_nc (table) -1, 1 + v * 3);
237 tab_text (table, 0, 1 + v * 3, TAB_LEFT, var_to_string (var));
238 tab_text (table, 1, 1 + v * 3, TAB_LEFT, _("Group1"));
239 tab_text (table, 1, 2 + v * 3, TAB_LEFT, _("Group2"));
240 tab_text (table, 1, 3 + v * 3, TAB_LEFT, _("Total"));
243 tab_double (table, 5, 1 + v * 3, TAB_NONE, bst->p, NULL);
245 /* Category labels */
246 tab_text (table, 2, 1 + v * 3, TAB_NONE, ds_cstr (&catstr1));
247 tab_text (table, 2, 2 + v * 3, TAB_NONE, ds_cstr (&catstr2));
250 tab_double (table, 3, 1 + v * 3, TAB_NONE, cat1[v].count, wfmt);
251 tab_double (table, 3, 2 + v * 3, TAB_NONE, cat2[v].count, wfmt);
253 n_total = cat1[v].count + cat2[v].count;
254 tab_double (table, 3, 3 + v * 3, TAB_NONE, n_total, wfmt);
256 /* Observed Proportions */
257 tab_double (table, 4, 1 + v * 3, TAB_NONE,
258 cat1[v].count / n_total, NULL);
259 tab_double (table, 4, 2 + v * 3, TAB_NONE,
260 cat2[v].count / n_total, NULL);
262 tab_double (table, 4, 3 + v * 3, TAB_NONE,
263 (cat1[v].count + cat2[v].count) / n_total, NULL);
266 sig = calculate_binomial (cat1[v].count, cat2[v].count, bst->p);
267 tab_double (table, 6, 1 + v * 3, TAB_NONE, sig, NULL);
269 ds_destroy (&catstr1);
270 ds_destroy (&catstr2);
273 tab_text (table, 2, 0, TAB_CENTER, _("Category"));
274 tab_text (table, 3, 0, TAB_CENTER, _("N"));
275 tab_text (table, 4, 0, TAB_CENTER, _("Observed Prop."));
276 tab_text (table, 5, 0, TAB_CENTER, _("Test Prop."));
278 tab_text (table, 6, 0, TAB_CENTER | TAT_PRINTF,
279 _("Exact Sig. (%d-tailed)"),
280 bst->p == 0.5 ? 2: 1);
282 tab_vline (table, TAL_2, 2, 0, tab_nr (table) -1);
286 for (v = 0; v < ost->n_vars; v++)
288 free (cat1[v].value);
289 free (cat2[v].value);