/* PSPP - a program for statistical analysis. -*-c-*-
- Copyright (C) 2006, 2008, 2009, 2010, 2011 Free Software Foundation, Inc.
+ Copyright (C) 2006, 2008, 2009, 2010, 2011, 2016 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
#include "language/lexer/variable-parser.h"
#include "language/stats/binomial.h"
#include "language/stats/chisquare.h"
+#include "language/stats/ks-one-sample.h"
#include "language/stats/cochran.h"
#include "language/stats/friedman.h"
+#include "language/stats/jonckheere-terpstra.h"
#include "language/stats/kruskal-wallis.h"
#include "language/stats/mann-whitney.h"
+#include "language/stats/mcnemar.h"
+#include "language/stats/median.h"
#include "language/stats/npar-summary.h"
#include "language/stats/runs.h"
#include "language/stats/sign.h"
int chisquare;
int cochran;
int binomial;
+ int ks_one_sample;
int wilcoxon;
int sign;
int runs;
int kendall;
int kruskal_wallis;
int mann_whitney;
+ int mcnemar;
+ int median;
+ int jonckheere_terpstra;
int missing;
int method;
int statistics;
/* Prototype for custom subcommands of NPAR TESTS. */
static int npar_chisquare (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_binomial (struct lexer *, struct dataset *, struct npar_specs *);
+static int npar_ks_one_sample (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_runs (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_friedman (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_kendall (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_wilcoxon (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_sign (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_kruskal_wallis (struct lexer *, struct dataset *, struct npar_specs *);
+static int npar_jonckheere_terpstra (struct lexer *, struct dataset *, struct npar_specs *);
static int npar_mann_whitney (struct lexer *, struct dataset *, struct npar_specs *);
+static int npar_mcnemar (struct lexer *, struct dataset *, struct npar_specs *);
+static int npar_median (struct lexer *, struct dataset *, struct npar_specs *);
+
static int npar_method (struct lexer *, struct npar_specs *);
/* Command parsing functions. */
parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests *npt,
struct npar_specs *nps)
{
- npt->binomial = 0;
npt->chisquare = 0;
npt->cochran = 0;
+ npt->binomial = 0;
+ npt->ks_one_sample = 0;
+ npt->wilcoxon = 0;
+ npt->sign = 0;
+ npt->runs = 0;
npt->friedman = 0;
+ npt->kendall = 0;
npt->kruskal_wallis = 0;
npt->mann_whitney = 0;
- npt->runs = 0;
- npt->sign = 0;
- npt->wilcoxon = 0;
- npt->missing = 0;
+ npt->mcnemar = 0;
+ npt->median = 0;
+ npt->jonckheere_terpstra = 0;
+
npt->miss = MISS_ANALYSIS;
+ npt->missing = 0;
npt->method = 0;
npt->statistics = 0;
+
memset (npt->a_statistics, 0, sizeof npt->a_statistics);
for (;;)
{
NOT_REACHED ();
}
}
+ else if (lex_match_phrase (lexer, "K-S") ||
+ lex_match_phrase (lexer, "KOLMOGOROV-SMIRNOV"))
+ {
+ lex_match (lexer, T_EQUALS);
+ npt->ks_one_sample++;
+ switch (npar_ks_one_sample (lexer, ds, nps))
+ {
+ case 0:
+ goto lossage;
+ case 1:
+ break;
+ case 2:
+ lex_error (lexer, NULL);
+ goto lossage;
+ default:
+ NOT_REACHED ();
+ }
+ }
+ else if (lex_match_phrase (lexer, "J-T") ||
+ lex_match_phrase (lexer, "JONCKHEERE-TERPSTRA"))
+ {
+ lex_match (lexer, T_EQUALS);
+ npt->jonckheere_terpstra++;
+ switch (npar_jonckheere_terpstra (lexer, ds, nps))
+ {
+ case 0:
+ goto lossage;
+ case 1:
+ break;
+ case 2:
+ lex_error (lexer, NULL);
+ goto lossage;
+ default:
+ NOT_REACHED ();
+ }
+ }
else if (lex_match_phrase (lexer, "K-W") ||
lex_match_phrase (lexer, "KRUSKAL-WALLIS"))
{
NOT_REACHED ();
}
}
+ else if (lex_match_phrase (lexer, "MCNEMAR"))
+ {
+ lex_match (lexer, T_EQUALS);
+ npt->mcnemar++;
+ switch (npar_mcnemar (lexer, ds, nps))
+ {
+ case 0:
+ goto lossage;
+ case 1:
+ break;
+ case 2:
+ lex_error (lexer, NULL);
+ goto lossage;
+ default:
+ NOT_REACHED ();
+ }
+ }
else if (lex_match_phrase (lexer, "M-W") ||
lex_match_phrase (lexer, "MANN-WHITNEY"))
{
default:
NOT_REACHED ();
}
+ }
+ else if (lex_match_phrase (lexer, "MEDIAN"))
+ {
+ npt->median++;
+
+ switch (npar_median (lexer, ds, nps))
+ {
+ case 0:
+ goto lossage;
+ case 1:
+ break;
+ case 2:
+ lex_error (lexer, NULL);
+ goto lossage;
+ default:
+ NOT_REACHED ();
+ }
}
else if (lex_match_id (lexer, "WILCOXON"))
{
npt->missing++;
if (npt->missing > 1)
{
- msg (SE, _("The %s subcommand may be given only once."), "MISSING");
+ lex_sbc_only_once ("MISSING");
goto lossage;
}
while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
npt->method++;
if (npt->method > 1)
{
- msg (SE, _("The %s subcommand may be given only once."), "METHOD");
+ lex_sbc_only_once ("METHOD");
goto lossage;
}
switch (npar_method (lexer, nps))
if (!lex_match (lexer, T_SLASH))
break;
}
-
if (lex_token (lexer) != T_ENDCMD)
{
lex_error (lexer, _("expecting end of command"));
const struct npar_test *test = specs->test[t];
if ( NULL == test->execute )
{
- msg (SW, _("NPAR subcommand not currently implemented."));
+ msg (SW, _("%s subcommand not currently implemented."), "NPAR");
continue;
}
test->execute (ds, casereader_clone (input), specs->filter, test, specs->exact, specs->timer);
}
- if ( specs->descriptives )
+ if (specs->descriptives && specs->n_vars > 0)
{
summary_descriptives = xnmalloc (sizeof (*summary_descriptives),
specs->n_vars);
if ( (specs->descriptives || specs->quartiles)
&& !taint_has_tainted_successor (casereader_get_taint (input)) )
- do_summary_box (summary_descriptives, specs->vv, specs->n_vars );
+ do_summary_box (summary_descriptives, specs->vv, specs->n_vars,
+ dict_get_weight_format (dataset_dict (ds)));
free (summary_descriptives);
casereader_destroy (input);
}
}
- sort (npar_specs.vv, npar_specs.n_vars, sizeof (*npar_specs.vv),
+ sort (npar_specs.vv, npar_specs.n_vars, sizeof (*npar_specs.vv),
compare_var_ptrs_by_name, NULL);
if ( cmd.statistics )
}
else
{
- lex_error (lexer, _("Expecting MEAN, MEDIAN, MODE or number"));
+ lex_error (lexer, _("Expecting %s, %s, %s or a number."), "MEAN", "MEDIAN", "MODE");
return 0;
}
-
- lex_force_match (lexer, T_RPAREN);
- lex_force_match (lexer, T_EQUALS);
+
+ if (! lex_force_match (lexer, T_RPAREN))
+ return 2;
+
+ if (! lex_force_match (lexer, T_EQUALS))
+ return 2;
+
if (!parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
&tp->vars, &tp->n_vars,
PV_NO_SCRATCH | PV_NO_DUPLICATE | PV_NUMERIC))
npar_friedman (struct lexer *lexer, struct dataset *ds,
struct npar_specs *specs)
{
- struct friedman_test *ft = pool_alloc (specs->pool, sizeof (*ft));
+ struct friedman_test *ft = pool_alloc (specs->pool, sizeof (*ft));
struct one_sample_test *ost = &ft->parent;
struct npar_test *nt = &ost->parent;
npar_kendall (struct lexer *lexer, struct dataset *ds,
struct npar_specs *specs)
{
- struct friedman_test *kt = pool_alloc (specs->pool, sizeof (*kt));
+ struct friedman_test *kt = pool_alloc (specs->pool, sizeof (*kt));
struct one_sample_test *ost = &kt->parent;
struct npar_test *nt = &ost->parent;
npar_cochran (struct lexer *lexer, struct dataset *ds,
struct npar_specs *specs)
{
- struct one_sample_test *ft = pool_alloc (specs->pool, sizeof (*ft));
+ struct one_sample_test *ft = pool_alloc (specs->pool, sizeof (*ft));
struct npar_test *nt = &ft->parent;
nt->execute = cochran_execute;
{
cstp->ranged = true;
if ( ! lex_force_num (lexer)) return 0;
- cstp->lo = lex_integer (lexer);
+ cstp->lo = lex_number (lexer);
lex_get (lexer);
- lex_force_match (lexer, T_COMMA);
+ if (! lex_force_match (lexer, T_COMMA)) return 0;
if (! lex_force_num (lexer) ) return 0;
- cstp->hi = lex_integer (lexer);
+ cstp->hi = lex_number (lexer);
if ( cstp->lo >= cstp->hi )
{
msg (ME,
cstp->expected = NULL;
if (lex_match_phrase (lexer, "/EXPECTED"))
{
- lex_force_match (lexer, T_EQUALS);
- if ( ! lex_match_id (lexer, "EQUAL") )
+ if (! lex_force_match (lexer, T_EQUALS)) return 0;
+ if (! lex_match_id (lexer, "EQUAL") )
{
double f;
int n;
if ( lex_match (lexer, T_ASTERISK))
{
n = f;
+ if (!lex_force_num (lexer))
+ return 0;
f = lex_number (lexer);
lex_get (lexer);
}
{
btp->p = lex_number (lexer);
lex_get (lexer);
- lex_force_match (lexer, T_RPAREN);
+ if (!lex_force_match (lexer, T_RPAREN))
+ return 0;
}
else
return 0;
else
equals = true;
- if (equals || lex_match (lexer, T_EQUALS) )
+ if (!equals)
+ if (!lex_force_match (lexer, T_EQUALS))
+ return 0;
+
+ {
+ if (parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
+ &tp->vars, &tp->n_vars,
+ PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
+ {
+ if (lex_match (lexer, T_LPAREN))
+ {
+ if (! lex_force_num (lexer))
+ return 2;
+ btp->category1 = lex_number (lexer);
+ lex_get (lexer);
+ if ( lex_match (lexer, T_COMMA))
+ {
+ if ( ! lex_force_num (lexer) ) return 2;
+ btp->category2 = lex_number (lexer);
+ lex_get (lexer);
+ }
+ else
+ {
+ btp->cutpoint = btp->category1;
+ }
+
+ if (! lex_force_match (lexer, T_RPAREN))
+ return 0;
+ }
+ }
+ else
+ {
+ return 2;
+ }
+ }
+
+ specs->n_tests++;
+ specs->test = pool_realloc (specs->pool,
+ specs->test,
+ sizeof (*specs->test) * specs->n_tests);
+
+ specs->test[specs->n_tests - 1] = nt;
+
+ return 1;
+}
+
+
+
+static void
+ks_one_sample_parse_params (struct lexer *lexer, struct ks_one_sample_test *kst, int params)
+{
+ assert (params == 1 || params == 2);
+
+ if (lex_is_number (lexer))
{
- if (parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
- &tp->vars, &tp->n_vars,
- PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
+ kst->p[0] = lex_number (lexer);
+
+ lex_get (lexer);
+ if ( params == 2)
{
- if (lex_match (lexer, T_LPAREN))
+ lex_match (lexer, T_COMMA);
+ if (lex_force_num (lexer))
{
- lex_force_num (lexer);
- btp->category1 = lex_number (lexer);
- lex_get (lexer);
- if ( lex_match (lexer, T_COMMA))
- {
- if ( ! lex_force_num (lexer) ) return 2;
- btp->category2 = lex_number (lexer);
- lex_get (lexer);
- }
- else
- {
- btp->cutpoint = btp->category1;
- }
-
- lex_force_match (lexer, T_RPAREN);
+ kst->p[1] = lex_number (lexer);
+ lex_get (lexer);
}
}
- else
- return 2;
+ }
+}
+
+static int
+npar_ks_one_sample (struct lexer *lexer, struct dataset *ds, struct npar_specs *specs)
+{
+ struct ks_one_sample_test *kst = pool_alloc (specs->pool, sizeof (*kst));
+ struct one_sample_test *tp = &kst->parent;
+ struct npar_test *nt = &tp->parent;
+ nt->execute = ks_one_sample_execute;
+ nt->insert_variables = one_sample_insert_variables;
+
+ kst->p[0] = kst->p[1] = SYSMIS;
+
+ if (! lex_force_match (lexer, T_LPAREN))
+ return 2;
+
+ if (lex_match_id (lexer, "NORMAL"))
+ {
+ kst->dist = KS_NORMAL;
+ ks_one_sample_parse_params (lexer, kst, 2);
+ }
+ else if (lex_match_id (lexer, "POISSON"))
+ {
+ kst->dist = KS_POISSON;
+ ks_one_sample_parse_params (lexer, kst, 1);
}
+ else if (lex_match_id (lexer, "UNIFORM"))
+ {
+ kst->dist = KS_UNIFORM;
+ ks_one_sample_parse_params (lexer, kst, 2);
+ }
+ else if (lex_match_id (lexer, "EXPONENTIAL"))
+ {
+ kst->dist = KS_EXPONENTIAL;
+ ks_one_sample_parse_params (lexer, kst, 1);
+ }
+ else
+ return 2;
+
+ if (! lex_force_match (lexer, T_RPAREN))
+ return 2;
+
+ lex_match (lexer, T_EQUALS);
+
+ if (! parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
+ &tp->vars, &tp->n_vars,
+ PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
+ return 2;
specs->n_tests++;
specs->test = pool_realloc (specs->pool,
}
-static bool
-parse_two_sample_related_test (struct lexer *lexer,
- const struct dictionary *dict,
- struct two_sample_test *test_parameters,
- struct pool *pool
- );
-
-
static bool
parse_two_sample_related_test (struct lexer *lexer,
const struct dictionary *dict,
struct two_sample_test *test_parameters,
- struct pool *pool
- )
+ struct pool *pool)
{
int n = 0;
bool paired = false;
if (!parse_variables_const_pool (lexer, pool,
dict,
&vlist1, &n_vlist1,
- PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
+ PV_NUMERIC | PV_NO_SCRATCH | PV_DUPLICATE) )
return false;
if ( lex_match (lexer, T_WITH))
with = true;
if ( !parse_variables_const_pool (lexer, pool, dict,
&vlist2, &n_vlist2,
- PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
+ PV_NUMERIC | PV_NO_SCRATCH | PV_DUPLICATE) )
return false;
paired = (lex_match (lexer, T_LPAREN) &&
if (paired)
{
if ( n_vlist1 != n_vlist2)
- msg (SE, _("PAIRED was specified but the number of variables "
+ {
+ msg (SE, _("PAIRED was specified but the number of variables "
"preceding WITH (%zu) did not match the number "
"following (%zu)."), n_vlist1, n_vlist2);
+ return false;
+ }
test_parameters->n_pairs = n_vlist1 ;
}
assert (n_vlist1 == n_vlist2);
for ( i = 0 ; i < n_vlist1; ++i )
{
- test_parameters->pairs[n][1] = vlist1[i];
- test_parameters->pairs[n][0] = vlist2[i];
+ test_parameters->pairs[n][0] = vlist1[i];
+ test_parameters->pairs[n][1] = vlist2[i];
n++;
}
}
{
for ( j = 0 ; j < n_vlist2; ++j )
{
- test_parameters->pairs[n][1] = vlist1[i];
- test_parameters->pairs[n][0] = vlist2[j];
+ test_parameters->pairs[n][0] = vlist1[i];
+ test_parameters->pairs[n][1] = vlist2[j];
n++;
}
}
for ( j = i + 1 ; j < n_vlist1; ++j )
{
assert ( n < test_parameters->n_pairs);
- test_parameters->pairs[n][1] = vlist1[i];
- test_parameters->pairs[n][0] = vlist1[j];
+ test_parameters->pairs[n][0] = vlist1[i];
+ test_parameters->pairs[n][1] = vlist1[j];
n++;
}
}
return false;
nst->indep_var = parse_variable_const (lexer, dict);
+ if (!nst->indep_var)
+ return false;
if ( ! lex_force_match (lexer, T_LPAREN))
return false;
struct dataset *ds,
struct npar_specs *specs )
{
-
-
struct two_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
struct npar_test *nt = &tp->parent;
nt->execute = wilcoxon_execute;
}
+static int
+npar_median (struct lexer *lexer,
+ struct dataset *ds,
+ struct npar_specs *specs)
+{
+ struct median_test *mt = pool_alloc (specs->pool, sizeof (*mt));
+ struct n_sample_test *tp = &mt->parent;
+ struct npar_test *nt = &tp->parent;
+
+ mt->median = SYSMIS;
+
+ if ( lex_match (lexer, T_LPAREN) && lex_force_num (lexer))
+ {
+ mt->median = lex_number (lexer);
+ lex_get (lexer);
+ if (! lex_force_match (lexer, T_RPAREN))
+ return 0;
+ }
+
+ lex_match (lexer, T_EQUALS);
+
+ nt->insert_variables = n_sample_insert_variables;
+ nt->execute = median_execute;
+
+ if (!parse_n_sample_related_test (lexer, dataset_dict (ds),
+ tp, specs->pool) )
+ return 0;
+
+ specs->n_tests++;
+ specs->test = pool_realloc (specs->pool,
+ specs->test,
+ sizeof (*specs->test) * specs->n_tests);
+ specs->test[specs->n_tests - 1] = nt;
+
+ return 1;
+}
+
+
static int
npar_sign (struct lexer *lexer, struct dataset *ds,
struct npar_specs *specs)
return 1;
}
+
+static int
+npar_mcnemar (struct lexer *lexer, struct dataset *ds,
+ struct npar_specs *specs)
+{
+ struct two_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
+ struct npar_test *nt = &tp->parent;
+
+ nt->execute = mcnemar_execute;
+
+ if (!parse_two_sample_related_test (lexer, dataset_dict (ds),
+ tp, specs->pool) )
+ return 0;
+
+ specs->n_tests++;
+ specs->test = pool_realloc (specs->pool,
+ specs->test,
+ sizeof (*specs->test) * specs->n_tests);
+ specs->test[specs->n_tests - 1] = nt;
+
+ return 1;
+}
+
+
+static int
+npar_jonckheere_terpstra (struct lexer *lexer, struct dataset *ds,
+ struct npar_specs *specs)
+{
+ struct n_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
+ struct npar_test *nt = &tp->parent;
+
+ nt->insert_variables = n_sample_insert_variables;
+
+ nt->execute = jonckheere_terpstra_execute;
+
+ if (!parse_n_sample_related_test (lexer, dataset_dict (ds),
+ tp, specs->pool) )
+ return 0;
+
+ specs->n_tests++;
+ specs->test = pool_realloc (specs->pool,
+ specs->test,
+ sizeof (*specs->test) * specs->n_tests);
+ specs->test[specs->n_tests - 1] = nt;
+
+ return 1;
+}
+
static int
npar_kruskal_wallis (struct lexer *lexer, struct dataset *ds,
struct npar_specs *specs)
size_t hash = hash_pointer (var, 0);
struct hmapx_node *node;
const struct variable *v = NULL;
-
+
HMAPX_FOR_EACH_WITH_HASH (v, node, hash, var_map)
{
if ( v == var)
}
}
-static void
+static void
n_sample_insert_variables (const struct npar_test *test,
struct hmapx *var_map)
{
if ( lex_match (lexer, T_LPAREN))
{
- if ( lex_force_num (lexer) )
+ if (lex_force_num (lexer) )
{
specs->timer = lex_number (lexer);
lex_get (lexer);
}
- lex_force_match (lexer, T_RPAREN);
+ if (lex_force_match (lexer, T_RPAREN))
+ return 0;
}
}
}