X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fchisquare.c;h=6cb7fc0b192ee2bfce5e50a620a26bd466a3ee6e;hb=14aac9fe7a7efbb6c9bded2ed5969a643cb76645;hp=337203f18deff6112f11ec91efb23f1e4b79b35a;hpb=9f087e7aa4cdff1d5d46d5e188c0017a9d2d0029;p=pspp-builds.git diff --git a/src/language/stats/chisquare.c b/src/language/stats/chisquare.c index 337203f1..6cb7fc0b 100644 --- a/src/language/stats/chisquare.c +++ b/src/language/stats/chisquare.c @@ -1,5 +1,5 @@ /* PSPP - a program for statistical analysis. - Copyright (C) 2006, 2007 Free Software Foundation, Inc. + Copyright (C) 2006, 2007, 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 @@ -21,6 +21,7 @@ #include #include +#include #include #include #include @@ -57,7 +58,7 @@ create_freq_hash_with_range (const struct dictionary *dict, { bool warn = true; float i_d; - struct ccase c; + struct ccase *c; struct hsh_table *freq_hash = hsh_create (4, compare_freq, hash_freq, @@ -78,21 +79,21 @@ create_freq_hash_with_range (const struct dictionary *dict, hsh_insert (freq_hash, fr); } - while (casereader_read (input, &c)) + while ((c = casereader_read (input)) != NULL) { union value obs_value; struct freq **existing_fr; struct freq *fr = xmalloc(sizeof (*fr)); - fr->value = case_data (&c, var); + fr->value = case_data (c, var); - fr->count = dict_get_case_weight (dict, &c, &warn); + fr->count = dict_get_case_weight (dict, c, &warn); obs_value.f = trunc (fr->value->f); if ( obs_value.f < lo || obs_value.f > hi) { free (fr); - case_destroy (&c); + case_unref (c); continue; } @@ -107,7 +108,7 @@ create_freq_hash_with_range (const struct dictionary *dict, (*existing_fr)->count += fr->count; free (fr); - case_destroy (&c); + case_unref (c); } if (casereader_destroy (input)) return freq_hash; @@ -130,20 +131,20 @@ create_freq_hash (const struct dictionary *dict, const struct variable *var) { bool warn = true; - struct ccase c; + struct ccase *c; struct hsh_table *freq_hash = hsh_create (4, compare_freq, hash_freq, free_freq_mutable_hash, (void *) var); - for (; casereader_read (input, &c); case_destroy (&c)) + for (; (c = casereader_read (input)) != NULL; case_unref (c)) { struct freq **existing_fr; struct freq *fr = xmalloc(sizeof (*fr)); - fr->value = case_data (&c, var); + fr->value = case_data (c, var); - fr->count = dict_get_case_weight (dict, &c, &warn); + fr->count = dict_get_case_weight (dict, c, &warn); existing_fr = (struct freq **) hsh_probe (freq_hash, fr); if ( *existing_fr) @@ -269,8 +270,8 @@ create_combo_frequency_table (const struct chisquare_test *test) } for ( i = test->lo ; i <= test->hi ; ++i ) - tab_float (table, 0, 2 + i - test->lo, - TAB_LEFT, 1 + i - test->lo, 8, 0); + tab_fixed (table, 0, 2 + i - test->lo, + TAB_LEFT, 1 + i - test->lo, 8, 0); tab_headers (table, 1, 0, 2, 0); @@ -320,7 +321,9 @@ void chisquare_execute (const struct dataset *ds, struct casereader *input, enum mv_class exclude, - const struct npar_test *test) + const struct npar_test *test, + bool exact UNUSED, + double timer UNUSED) { const struct dictionary *dict = dataset_dict (ds); int v, i; @@ -328,6 +331,9 @@ chisquare_execute (const struct dataset *ds, struct chisquare_test *cst = (struct chisquare_test *) test; int n_cells = 0; double total_expected = 0.0; + const struct variable *wvar = dict_get_weight (dict); + const struct fmt_spec *wfmt = wvar ? + var_get_print_format (wvar) : & F_8_0; double *df = xzalloc (sizeof (*df) * ost->n_vars); double *xsq = xzalloc (sizeof (*df) * ost->n_vars); @@ -344,7 +350,8 @@ chisquare_execute (const struct dataset *ds, struct hsh_table *freq_hash = NULL; struct casereader *reader = casereader_create_filter_missing (casereader_clone (input), - &ost->vars[v], 1, exclude, NULL); + &ost->vars[v], 1, exclude, + NULL, NULL); struct tab_table *freq_table = create_variable_frequency_table(dict, reader, cst, v, &freq_hash); @@ -362,36 +369,41 @@ chisquare_execute (const struct dataset *ds, xsq[v] = 0.0; for ( i = 0 ; i < n_cells ; ++i ) { + struct string str; double exp; const union value *observed_value = ff[i]->value; + ds_init_empty (&str); + var_append_value_name (ost->vars[v], observed_value, &str); + /* The key */ - tab_text (freq_table, 0, i + 1, TAB_LEFT, - var_get_value_name (ost->vars[v], observed_value)); + tab_text (freq_table, 0, i + 1, TAB_LEFT, ds_cstr (&str)); + ds_destroy (&str); + /* The observed N */ - tab_float (freq_table, 1, i + 1, TAB_NONE, - ff[i]->count, 8, 0); + tab_double (freq_table, 1, i + 1, TAB_NONE, + ff[i]->count, wfmt); if ( cst->n_expected > 0 ) exp = cst->expected[i] * total_obs / total_expected ; else exp = total_obs / (double) n_cells; - tab_float (freq_table, 2, i + 1, TAB_NONE, - exp, 8, 2); + tab_double (freq_table, 2, i + 1, TAB_NONE, + exp, NULL); /* The residual */ - tab_float (freq_table, 3, i + 1, TAB_NONE, - ff[i]->count - exp, 8, 2); + tab_double (freq_table, 3, i + 1, TAB_NONE, + ff[i]->count - exp, NULL); xsq[v] += (ff[i]->count - exp) * (ff[i]->count - exp) / exp; } df[v] = n_cells - 1.0; - tab_float (freq_table, 1, i + 1, TAB_NONE, - total_obs, 8, 0); + tab_double (freq_table, 1, i + 1, TAB_NONE, + total_obs, wfmt); tab_submit (freq_table); @@ -409,7 +421,8 @@ chisquare_execute (const struct dataset *ds, double total_obs = 0.0; struct casereader *reader = casereader_create_filter_missing (casereader_clone (input), - &ost->vars[v], 1, exclude, NULL); + &ost->vars[v], 1, exclude, + NULL, NULL); struct hsh_table *freq_hash = create_freq_hash_with_range (dict, reader, ost->vars[v], cst->lo, cst->hi); @@ -428,17 +441,21 @@ chisquare_execute (const struct dataset *ds, xsq[v] = 0.0; for ( i = 0 ; i < hsh_count (freq_hash) ; ++i ) { + struct string str; double exp; const union value *observed_value = ff[i]->value; + ds_init_empty (&str); + var_append_value_name (ost->vars[v], observed_value, &str); /* The key */ tab_text (freq_table, v * 4 + 1, i + 2 , TAB_LEFT, - var_get_value_name (ost->vars[v], observed_value)); + ds_cstr (&str)); + ds_destroy (&str); /* The observed N */ - tab_float (freq_table, v * 4 + 2, i + 2 , TAB_NONE, - ff[i]->count, 8, 0); + tab_double (freq_table, v * 4 + 2, i + 2 , TAB_NONE, + ff[i]->count, wfmt); if ( cst->n_expected > 0 ) exp = cst->expected[i] * total_obs / total_expected ; @@ -446,19 +463,19 @@ chisquare_execute (const struct dataset *ds, exp = total_obs / (double) hsh_count (freq_hash); /* The expected N */ - tab_float (freq_table, v * 4 + 3, i + 2 , TAB_NONE, - exp, 8, 2); + tab_double (freq_table, v * 4 + 3, i + 2 , TAB_NONE, + exp, NULL); /* The residual */ - tab_float (freq_table, v * 4 + 4, i + 2 , TAB_NONE, - ff[i]->count - exp, 8, 2); + tab_double (freq_table, v * 4 + 4, i + 2 , TAB_NONE, + ff[i]->count - exp, NULL); xsq[v] += (ff[i]->count - exp) * (ff[i]->count - exp) / exp; } - tab_float (freq_table, v * 4 + 2, tab_nr (freq_table) - 1, TAB_NONE, - total_obs, 8, 0); + tab_double (freq_table, v * 4 + 2, tab_nr (freq_table) - 1, TAB_NONE, + total_obs, wfmt); df[v] = n_cells - 1.0; @@ -481,11 +498,11 @@ chisquare_execute (const struct dataset *ds, tab_text (stats_table, 1 + v, 0, TAB_CENTER, var_get_name (var)); - tab_float (stats_table, 1 + v, 1, TAB_NONE, xsq[v], 8,3); - tab_float (stats_table, 1 + v, 2, TAB_NONE, df[v], 8,0); + tab_double (stats_table, 1 + v, 1, TAB_NONE, xsq[v], NULL); + tab_fixed (stats_table, 1 + v, 2, TAB_NONE, df[v], 8, 0); - tab_float (stats_table, 1 + v, 3, TAB_NONE, - gsl_cdf_chisq_Q (xsq[v], df[v]), 8,3); + tab_double (stats_table, 1 + v, 3, TAB_NONE, + gsl_cdf_chisq_Q (xsq[v], df[v]), NULL); } tab_submit (stats_table); }