X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fdescriptives.c;h=6afb6b53a4f336d29e72fa2b0fffdefec7ded38c;hb=480a0746507ce73d26f528b56dc3ed80195096e0;hp=6704fe5d2408db4ff920e3f82e2c38e1ae3d7d63;hpb=2322678e8fddbbf158b01b2720db2636404bba3b;p=pspp diff --git a/src/language/stats/descriptives.c b/src/language/stats/descriptives.c index 6704fe5d24..6afb6b53a4 100644 --- a/src/language/stats/descriptives.c +++ b/src/language/stats/descriptives.c @@ -1,6 +1,5 @@ /* PSPP - computes sample statistics. Copyright (C) 1997-9, 2000 Free Software Foundation, Inc. - Written by Ben Pfaff . This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as @@ -20,25 +19,30 @@ /* FIXME: Many possible optimizations. */ #include -#include "message.h" + #include #include #include -#include "array.h" -#include "alloc.h" -#include "case.h" -#include "casefile.h" -#include "command.h" -#include "compiler.h" -#include "dictionary.h" -#include "lexer.h" -#include "message.h" -#include "magic.h" -#include "moments.h" -#include "manager.h" -#include "table.h" -#include "variable.h" -#include "procedure.h" + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include #include "gettext.h" #define _(msgid) gettext (msgid) @@ -59,11 +63,10 @@ enum dsc_missing_type calculating a Z-score. */ struct dsc_z_score { - int src_idx; /* Source index into case data. */ - int dst_idx; /* Destination index into case data. */ + struct variable *src_var; /* Variable on which z-score is based. */ + struct variable *z_var; /* New z-score variable. */ double mean; /* Distribution mean. */ double std_dev; /* Distribution standard deviation. */ - struct variable *v; /* Variable on which z-score is based. */ }; /* DESCRIPTIVES transformation (for calculating Z-scores). */ @@ -155,7 +158,7 @@ struct dsc_proc /* Accumulated results. */ double missing_listwise; /* Sum of weights of cases missing listwise. */ double valid; /* Sum of weights of valid cases. */ - int bad_warn; /* Warn if bad weight found. */ + bool bad_warn; /* Warn if bad weight found. */ enum dsc_statistic sort_by_stat; /* Statistic to sort by; -1: name. */ int sort_ascending; /* !0: ascending order; 0: descending. */ unsigned long show_stats; /* Statistics to display. */ @@ -164,26 +167,31 @@ struct dsc_proc }; /* Parsing. */ -static enum dsc_statistic match_statistic (void); +static enum dsc_statistic match_statistic (struct lexer *); static void free_dsc_proc (struct dsc_proc *); /* Z-score functions. */ -static int try_name (struct dsc_proc *dsc, char *name); -static int generate_z_varname (struct dsc_proc *dsc, char *z_name, - const char *name, size_t *z_cnt); +static bool try_name (const struct dictionary *dict, + struct dsc_proc *dsc, const char *name); +static bool generate_z_varname (const struct dictionary *dict, + struct dsc_proc *dsc, char *z_name, + const char *name, size_t *z_cnt); static void dump_z_table (struct dsc_proc *); -static void setup_z_trns (struct dsc_proc *); +static void setup_z_trns (struct dsc_proc *, struct dataset *); /* Procedure execution functions. */ -static bool calc_descriptives (const struct casefile *, void *dsc_); +static bool calc_descriptives (const struct ccase *first, + const struct casefile *, void *dsc_, + const struct dataset *); static void display (struct dsc_proc *dsc); /* Parser and outline. */ /* Handles DESCRIPTIVES. */ int -cmd_descriptives (void) +cmd_descriptives (struct lexer *lexer, struct dataset *ds) { + struct dictionary *dict = dataset_dict (ds); struct dsc_proc *dsc; struct variable **vars = NULL; size_t var_cnt = 0; @@ -209,106 +217,106 @@ cmd_descriptives (void) dsc->show_stats = dsc->calc_stats = DEFAULT_STATS; /* Parse DESCRIPTIVES. */ - while (token != '.') + while (lex_token (lexer) != '.') { - if (lex_match_id ("MISSING")) + if (lex_match_id (lexer, "MISSING")) { - lex_match ('='); - while (token != '.' && token != '/') + lex_match (lexer, '='); + while (lex_token (lexer) != '.' && lex_token (lexer) != '/') { - if (lex_match_id ("VARIABLE")) + if (lex_match_id (lexer, "VARIABLE")) dsc->missing_type = DSC_VARIABLE; - else if (lex_match_id ("LISTWISE")) + else if (lex_match_id (lexer, "LISTWISE")) dsc->missing_type = DSC_LISTWISE; - else if (lex_match_id ("INCLUDE")) + else if (lex_match_id (lexer, "INCLUDE")) dsc->include_user_missing = 1; else { - lex_error (NULL); + lex_error (lexer, NULL); goto error; } - lex_match (','); + lex_match (lexer, ','); } } - else if (lex_match_id ("SAVE")) + else if (lex_match_id (lexer, "SAVE")) save_z_scores = 1; - else if (lex_match_id ("FORMAT")) + else if (lex_match_id (lexer, "FORMAT")) { - lex_match ('='); - while (token != '.' && token != '/') + lex_match (lexer, '='); + while (lex_token (lexer) != '.' && lex_token (lexer) != '/') { - if (lex_match_id ("LABELS")) + if (lex_match_id (lexer, "LABELS")) dsc->show_var_labels = 1; - else if (lex_match_id ("NOLABELS")) + else if (lex_match_id (lexer, "NOLABELS")) dsc->show_var_labels = 0; - else if (lex_match_id ("INDEX")) + else if (lex_match_id (lexer, "INDEX")) dsc->show_index = 1; - else if (lex_match_id ("NOINDEX")) + else if (lex_match_id (lexer, "NOINDEX")) dsc->show_index = 0; - else if (lex_match_id ("LINE")) + else if (lex_match_id (lexer, "LINE")) dsc->format = DSC_LINE; - else if (lex_match_id ("SERIAL")) + else if (lex_match_id (lexer, "SERIAL")) dsc->format = DSC_SERIAL; else { - lex_error (NULL); + lex_error (lexer, NULL); goto error; } - lex_match (','); + lex_match (lexer, ','); } } - else if (lex_match_id ("STATISTICS")) + else if (lex_match_id (lexer, "STATISTICS")) { - lex_match ('='); + lex_match (lexer, '='); dsc->show_stats = 0; - while (token != '.' && token != '/') + while (lex_token (lexer) != '.' && lex_token (lexer) != '/') { - if (lex_match (T_ALL)) + if (lex_match (lexer, T_ALL)) dsc->show_stats |= (1ul << DSC_N_STATS) - 1; - else if (lex_match_id ("DEFAULT")) + else if (lex_match_id (lexer, "DEFAULT")) dsc->show_stats |= DEFAULT_STATS; else - dsc->show_stats |= 1ul << (match_statistic ()); - lex_match (','); + dsc->show_stats |= 1ul << (match_statistic (lexer)); + lex_match (lexer, ','); } if (dsc->show_stats == 0) dsc->show_stats = DEFAULT_STATS; } - else if (lex_match_id ("SORT")) + else if (lex_match_id (lexer, "SORT")) { - lex_match ('='); - if (lex_match_id ("NAME")) + lex_match (lexer, '='); + if (lex_match_id (lexer, "NAME")) dsc->sort_by_stat = DSC_NAME; else { - dsc->sort_by_stat = match_statistic (); + dsc->sort_by_stat = match_statistic (lexer); if (dsc->sort_by_stat == DSC_NONE ) dsc->sort_by_stat = DSC_MEAN; } - if (lex_match ('(')) + if (lex_match (lexer, '(')) { - if (lex_match_id ("A")) + if (lex_match_id (lexer, "A")) dsc->sort_ascending = 1; - else if (lex_match_id ("D")) + else if (lex_match_id (lexer, "D")) dsc->sort_ascending = 0; else - lex_error (NULL); - lex_force_match (')'); + lex_error (lexer, NULL); + lex_force_match (lexer, ')'); } } else if (var_cnt == 0) { - if (lex_look_ahead () == '=') + if (lex_look_ahead (lexer) == '=') { - lex_match_id ("VARIABLES"); - lex_match ('='); + lex_match_id (lexer, "VARIABLES"); + lex_match (lexer, '='); } - while (token != '.' && token != '/') + while (lex_token (lexer) != '.' && lex_token (lexer) != '/') { int i; - if (!parse_variables (default_dict, &vars, &var_cnt, + if (!parse_variables (lexer, dataset_dict (ds), &vars, &var_cnt, PV_APPEND | PV_NO_DUPLICATE | PV_NUMERIC)) goto error; @@ -322,34 +330,34 @@ cmd_descriptives (void) } dsc->var_cnt = var_cnt; - if (lex_match ('(')) + if (lex_match (lexer, '(')) { - if (token != T_ID) + if (lex_token (lexer) != T_ID) { - lex_error (NULL); + lex_error (lexer, NULL); goto error; } - if (try_name (dsc, tokid)) + if (try_name (dict, dsc, lex_tokid (lexer))) { - strcpy (dsc->vars[dsc->var_cnt - 1].z_name, tokid); + strcpy (dsc->vars[dsc->var_cnt - 1].z_name, lex_tokid (lexer)); z_cnt++; } else msg (SE, _("Z-score variable name %s would be" - " a duplicate variable name."), tokid); - lex_get (); - if (!lex_force_match (')')) + " a duplicate variable name."), lex_tokid (lexer)); + lex_get (lexer); + if (!lex_force_match (lexer, ')')) goto error; } } } else { - lex_error (NULL); + lex_error (lexer, NULL); goto error; } - lex_match ('/'); + lex_match (lexer, '/'); } if (var_cnt == 0) { @@ -367,8 +375,9 @@ cmd_descriptives (void) for (i = 0; i < dsc->var_cnt; i++) if (dsc->vars[i].z_name[0] == 0) { - if (!generate_z_varname (dsc, dsc->vars[i].z_name, - dsc->vars[i].v->name, &gen_cnt)) + if (!generate_z_varname (dict, dsc, dsc->vars[i].z_name, + var_get_name (dsc->vars[i].v), + &gen_cnt)) goto error; z_cnt++; } @@ -404,11 +413,11 @@ cmd_descriptives (void) dsc->vars[i].moments = moments_create (dsc->max_moment); /* Data pass. */ - ok = multipass_procedure_with_splits (calc_descriptives, dsc); + ok = multipass_procedure_with_splits (ds, calc_descriptives, dsc); /* Z-scoring! */ if (ok && z_cnt) - setup_z_trns (dsc); + setup_z_trns (dsc, ds); /* Done. */ free (vars); @@ -426,18 +435,18 @@ cmd_descriptives (void) specifiers). Emits an error if the current token ID does not name a statistic. */ static enum dsc_statistic -match_statistic (void) +match_statistic (struct lexer *lexer) { - if (token == T_ID) + if (lex_token (lexer) == T_ID) { enum dsc_statistic stat; for (stat = 0; stat < DSC_N_STATS; stat++) - if (lex_match_id (dsc_info[stat].identifier)) + if (lex_match_id (lexer, dsc_info[stat].identifier)) return stat; - lex_get(); - lex_error (_("expecting statistic name: reverting to default")); + lex_get (lexer); + lex_error (lexer, _("expecting statistic name: reverting to default")); } return DSC_NONE; @@ -460,27 +469,28 @@ free_dsc_proc (struct dsc_proc *dsc) /* Z scores. */ -/* Returns 0 if NAME is a duplicate of any existing variable name or - of any previously-declared z-var name; otherwise returns 1. */ -static int -try_name (struct dsc_proc *dsc, char *name) +/* Returns false if NAME is a duplicate of any existing variable name or + of any previously-declared z-var name; otherwise returns true. */ +static bool +try_name (const struct dictionary *dict, struct dsc_proc *dsc, + const char *name) { size_t i; - if (dict_lookup_var (default_dict, name) != NULL) - return 0; + if (dict_lookup_var (dict, name) != NULL) + return false; for (i = 0; i < dsc->var_cnt; i++) if (!strcasecmp (dsc->vars[i].z_name, name)) - return 0; - return 1; + return false; + return true; } /* Generates a name for a Z-score variable based on a variable named VAR_NAME, given that *Z_CNT generated variable names are - known to already exist. If successful, returns nonzero and - copies the new name into Z_NAME. On failure, returns zero. */ -static int -generate_z_varname (struct dsc_proc *dsc, char *z_name, + known to already exist. If successful, returns true and + copies the new name into Z_NAME. On failure, returns false. */ +static bool +generate_z_varname (const struct dictionary *dict, struct dsc_proc *dsc, char *z_name, const char *var_name, size_t *z_cnt) { char name[LONG_NAME_LEN + 1]; @@ -488,10 +498,10 @@ generate_z_varname (struct dsc_proc *dsc, char *z_name, /* Try a name based on the original variable name. */ name[0] = 'Z'; str_copy_trunc (name + 1, sizeof name - 1, var_name); - if (try_name (dsc, name)) + if (try_name (dict, dsc, name)) { strcpy (z_name, name); - return 1; + return true; } /* Generate a synthetic name. */ @@ -512,15 +522,16 @@ generate_z_varname (struct dsc_proc *dsc, char *z_name, msg (SE, _("Ran out of generic names for Z-score variables. " "There are only 126 generic names: ZSC001-ZSC0999, " "STDZ01-STDZ09, ZZZZ01-ZZZZ09, ZQZQ01-ZQZQ09.")); - return 0; + return false; } - if (try_name (dsc, name)) + if (try_name (dict, dsc, name)) { strcpy (z_name, name); - return 1; + return true; } } + NOT_REACHED(); } /* Outputs a table describing the mapping between source @@ -540,7 +551,7 @@ dump_z_table (struct dsc_proc *dsc) } t = tab_create (2, cnt + 1, 0); - tab_title (t, 0, _("Mapping of variables to corresponding Z-scores.")); + tab_title (t, _("Mapping of variables to corresponding Z-scores.")); tab_columns (t, SOM_COL_DOWN, 1); tab_headers (t, 0, 0, 1, 0); tab_box (t, TAL_1, TAL_1, TAL_0, TAL_1, 0, 0, 1, cnt); @@ -555,7 +566,7 @@ dump_z_table (struct dsc_proc *dsc) for (i = 0, y = 1; i < dsc->var_cnt; i++) if (dsc->vars[i].z_name[0] != '\0') { - tab_text (t, 0, y, TAB_LEFT, dsc->vars[i].v->name); + tab_text (t, 0, y, TAB_LEFT, var_get_name (dsc->vars[i].v)); tab_text (t, 1, y++, TAB_LEFT, dsc->vars[i].z_name); } } @@ -571,7 +582,7 @@ dump_z_table (struct dsc_proc *dsc) */ static int descriptives_trns_proc (void *trns_, struct ccase * c, - int case_idx UNUSED) + casenumber case_idx UNUSED) { struct dsc_trns *t = trns_; struct dsc_z_score *z; @@ -583,10 +594,10 @@ descriptives_trns_proc (void *trns_, struct ccase * c, assert(t->vars); for (vars = t->vars; vars < t->vars + t->var_cnt; vars++) { - double score = case_num (c, (*vars)->fv); + double score = case_num (c, *vars); if ( score == SYSMIS || (!t->include_user_missing - && mv_is_num_user_missing (&(*vars)->miss, score))) + && var_is_num_user_missing (*vars, score))) { all_sysmis = 1; break; @@ -596,13 +607,13 @@ descriptives_trns_proc (void *trns_, struct ccase * c, for (z = t->z_scores; z < t->z_scores + t->z_score_cnt; z++) { - double input = case_num (c, z->src_idx); - double *output = &case_data_rw (c, z->dst_idx)->f; + double input = case_num (c, z->src_var); + double *output = &case_data_rw (c, z->z_var)->f; if (z->mean == SYSMIS || z->std_dev == SYSMIS || all_sysmis || input == SYSMIS || (!t->include_user_missing - && mv_is_num_user_missing (&z->v->miss, input))) + && var_is_num_user_missing (z->src_var, input))) *output = SYSMIS; else *output = (input - z->mean) / z->std_dev; @@ -624,7 +635,7 @@ descriptives_trns_free (void *trns_) /* Sets up a transformation to calculate Z scores. */ static void -setup_z_trns (struct dsc_proc *dsc) +setup_z_trns (struct dsc_proc *dsc, struct dataset *ds) { struct dsc_trns *t; size_t cnt, i; @@ -657,50 +668,42 @@ setup_z_trns (struct dsc_proc *dsc) if (dv->z_name[0] != '\0') { struct dsc_z_score *z; - char *cp; struct variable *dst_var; - dst_var = dict_create_var_assert (default_dict, dv->z_name, 0); - dst_var->init = 0; - if (dv->v->label) - { - dst_var->label = xmalloc (strlen (dv->v->label) + 12); - cp = stpcpy (dst_var->label, _("Z-score of ")); - strcpy (cp, dv->v->label); - } - else - { - dst_var->label = xmalloc (strlen (dv->v->name) + 12); - cp = stpcpy (dst_var->label, _("Z-score of ")); - strcpy (cp, dv->v->name); - } + dst_var = dict_create_var_assert (dataset_dict (ds), dv->z_name, 0); + var_set_label (dst_var, xasprintf (_("Z-score of %s"), + var_to_string (dv->v))); z = &t->z_scores[cnt++]; - z->src_idx = dv->v->fv; - z->dst_idx = dst_var->fv; + z->src_var = dv->v; + z->z_var = dst_var; z->mean = dv->stats[DSC_MEAN]; z->std_dev = dv->stats[DSC_STDDEV]; - z->v = dv->v; } } - add_transformation (descriptives_trns_proc, descriptives_trns_free, t); + add_transformation (ds, + descriptives_trns_proc, descriptives_trns_free, t); } /* Statistical calculation. */ -static int listwise_missing (struct dsc_proc *dsc, const struct ccase *c); +static bool listwise_missing (struct dsc_proc *dsc, const struct ccase *c); /* Calculates and displays descriptive statistics for the cases in CF. */ static bool -calc_descriptives (const struct casefile *cf, void *dsc_) +calc_descriptives (const struct ccase *first, + const struct casefile *cf, void *dsc_, + const struct dataset *ds) { struct dsc_proc *dsc = dsc_; struct casereader *reader; struct ccase c; size_t i; + output_split_file_values (ds, first); + for (i = 0; i < dsc->var_cnt; i++) { struct dsc_var *dv = &dsc->vars[i]; @@ -715,11 +718,11 @@ calc_descriptives (const struct casefile *cf, void *dsc_) dsc->valid = 0.; /* First pass to handle most of the work. */ - for (reader = casefile_get_reader (cf); + for (reader = casefile_get_reader (cf, NULL); casereader_read (reader, &c); case_destroy (&c)) { - double weight = dict_get_case_weight (default_dict, &c, &dsc->bad_warn); + double weight = dict_get_case_weight (dataset_dict (ds), &c, &dsc->bad_warn); if (weight <= 0.0) continue; @@ -735,12 +738,12 @@ calc_descriptives (const struct casefile *cf, void *dsc_) for (i = 0; i < dsc->var_cnt; i++) { struct dsc_var *dv = &dsc->vars[i]; - double x = case_num (&c, dv->v->fv); + double x = case_num (&c, dv->v); if (dsc->missing_type != DSC_LISTWISE && (x == SYSMIS || (!dsc->include_user_missing - && mv_is_num_user_missing (&dv->v->miss, x)))) + && var_is_num_user_missing (dv->v, x)))) { dv->missing += weight; continue; @@ -760,11 +763,11 @@ calc_descriptives (const struct casefile *cf, void *dsc_) /* Second pass for higher-order moments. */ if (dsc->max_moment > MOMENT_MEAN) { - for (reader = casefile_get_reader (cf); + for (reader = casefile_get_reader (cf, NULL); casereader_read (reader, &c); case_destroy (&c)) { - double weight = dict_get_case_weight (default_dict, &c, + double weight = dict_get_case_weight (dataset_dict (ds), &c, &dsc->bad_warn); if (weight <= 0.0) continue; @@ -777,12 +780,12 @@ calc_descriptives (const struct casefile *cf, void *dsc_) for (i = 0; i < dsc->var_cnt; i++) { struct dsc_var *dv = &dsc->vars[i]; - double x = case_num (&c, dv->v->fv); + double x = case_num (&c, dv->v); if (dsc->missing_type != DSC_LISTWISE && (x == SYSMIS || (!dsc->include_user_missing - && mv_is_num_user_missing (&dv->v->miss, x)))) + && var_is_num_user_missing (dv->v, x)))) continue; if (dv->moments != NULL) @@ -834,9 +837,9 @@ calc_descriptives (const struct casefile *cf, void *dsc_) return true; } -/* Returns nonzero if any of the descriptives variables in DSC's - variable list have missing values in case C, zero otherwise. */ -static int +/* Returns true if any of the descriptives variables in DSC's + variable list have missing values in case C, false otherwise. */ +static bool listwise_missing (struct dsc_proc *dsc, const struct ccase *c) { size_t i; @@ -844,14 +847,14 @@ listwise_missing (struct dsc_proc *dsc, const struct ccase *c) for (i = 0; i < dsc->var_cnt; i++) { struct dsc_var *dv = &dsc->vars[i]; - double x = case_num (c, dv->v->fv); + double x = case_num (c, dv->v); if (x == SYSMIS || (!dsc->include_user_missing - && mv_is_num_user_missing (&dv->v->miss, x))) - return 1; + && var_is_num_user_missing (dv->v, x))) + return true; } - return 0; + return false; } /* Statistical display. */ @@ -906,7 +909,7 @@ display (struct dsc_proc *dsc) size_t j; nc = 0; - tab_text (t, nc++, i + 1, TAB_LEFT, dv->v->name); + tab_text (t, nc++, i + 1, TAB_LEFT, var_get_name (dv->v)); tab_text (t, nc++, i + 1, TAT_PRINTF, "%g", dv->valid); if (dsc->format == DSC_SERIAL) tab_text (t, nc++, i + 1, TAT_PRINTF, "%g", dv->missing); @@ -915,7 +918,7 @@ display (struct dsc_proc *dsc) tab_float (t, nc++, i + 1, TAB_NONE, dv->stats[j], 10, 3); } - tab_title (t, 1, _("Valid cases = %g; cases with missing value(s) = %g."), + tab_title (t, _("Valid cases = %g; cases with missing value(s) = %g."), dsc->valid, dsc->missing_listwise); tab_submit (t); @@ -924,16 +927,16 @@ display (struct dsc_proc *dsc) /* Compares `struct dsc_var's A and B according to the ordering specified by CMD. */ static int -descriptives_compare_dsc_vars (const void *a_, const void *b_, void *dsc_) +descriptives_compare_dsc_vars (const void *a_, const void *b_, const void *dsc_) { const struct dsc_var *a = a_; const struct dsc_var *b = b_; - struct dsc_proc *dsc = dsc_; + const struct dsc_proc *dsc = dsc_; int result; if (dsc->sort_by_stat == DSC_NAME) - result = strcasecmp (a->v->name, b->v->name); + result = strcasecmp (var_get_name (a->v), var_get_name (b->v)); else { double as = a->stats[dsc->sort_by_stat];