X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fnpar.c;h=5d8d8742f9c51935a32966c17b7722e3acb59344;hb=912833f9331784d692ade66c76bffb5497b890e0;hp=d34ec11932e5ffdd204b00f8a1fe850b3b873821;hpb=8ebbf609a32c5dc24043c02559749e07695ece2b;p=pspp diff --git a/src/language/stats/npar.c b/src/language/stats/npar.c index d34ec11932..5d8d8742f9 100644 --- a/src/language/stats/npar.c +++ b/src/language/stats/npar.c @@ -34,10 +34,14 @@ #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" @@ -82,6 +86,7 @@ struct cmd_npar_tests int chisquare; int cochran; int binomial; + int ks_one_sample; int wilcoxon; int sign; int runs; @@ -89,6 +94,9 @@ struct cmd_npar_tests int kendall; int kruskal_wallis; int mann_whitney; + int mcnemar; + int median; + int jonckheere_terpstra; int missing; int method; int statistics; @@ -124,6 +132,7 @@ struct npar_specs /* 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 *); @@ -131,7 +140,11 @@ static int npar_cochran (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. */ @@ -142,19 +155,26 @@ static int 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 (;;) { @@ -258,6 +278,42 @@ parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests 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")) { @@ -276,6 +332,23 @@ parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests 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")) { @@ -293,6 +366,23 @@ parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests 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")) { @@ -334,7 +424,7 @@ parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests 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) @@ -361,7 +451,7 @@ parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests 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)) @@ -444,7 +534,7 @@ npar_execute (struct casereader *input, 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); @@ -598,7 +688,7 @@ npar_runs (struct lexer *lexer, struct dataset *ds, } 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; } @@ -740,11 +830,11 @@ npar_chisquare (struct lexer *lexer, struct dataset *ds, { 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_num (lexer) ) return 0; - cstp->hi = lex_integer (lexer); + cstp->hi = lex_number (lexer); if ( cstp->lo >= cstp->hi ) { msg (ME, @@ -855,7 +945,8 @@ npar_binomial (struct lexer *lexer, struct dataset *ds, { if (lex_match (lexer, T_LPAREN)) { - lex_force_num (lexer); + if (! lex_force_num (lexer)) + return 2; btp->category1 = lex_number (lexer); lex_get (lexer); if ( lex_match (lexer, T_COMMA)) @@ -888,12 +979,93 @@ npar_binomial (struct lexer *lexer, struct dataset *ds, } + +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)) + { + kst->p[0] = lex_number (lexer); + + lex_get (lexer); + if ( params == 2) + { + lex_match (lexer, T_COMMA); + if (lex_force_num (lexer)) + { + kst->p[1] = lex_number (lexer); + lex_get (lexer); + } + } + } +} + +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, + specs->test, + sizeof (*specs->test) * specs->n_tests); + + specs->test[specs->n_tests - 1] = nt; + + return 1; +} + + 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; @@ -909,7 +1081,7 @@ parse_two_sample_related_test (struct lexer *lexer, 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)) @@ -917,7 +1089,7 @@ parse_two_sample_related_test (struct lexer *lexer, 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) && @@ -930,9 +1102,12 @@ parse_two_sample_related_test (struct lexer *lexer, 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 ; } @@ -1045,8 +1220,6 @@ npar_wilcoxon (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 = wilcoxon_execute; @@ -1090,6 +1263,43 @@ npar_mann_whitney (struct lexer *lexer, } +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); + lex_force_match (lexer, T_RPAREN); + } + + 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) @@ -1112,6 +1322,54 @@ npar_sign (struct lexer *lexer, struct dataset *ds, 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)