X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fdescriptives.c;h=54bc49d946a064bd3c4eee8ffdab962d611733c3;hb=2c0ddcd27c6e5afa944f9eca6f5cf6e7f6e32f98;hp=16869a1829651136b4ccde671849bc4f3b4ce27d;hpb=ab249f50e9f994b9a92a74b148f596f47fee0de9;p=pspp diff --git a/src/language/stats/descriptives.c b/src/language/stats/descriptives.c index 16869a1829..54bc49d946 100644 --- a/src/language/stats/descriptives.c +++ b/src/language/stats/descriptives.c @@ -1,5 +1,5 @@ /* PSPP - a program for statistical analysis. - Copyright (C) 1997-9, 2000, 2009 Free Software Foundation, Inc. + Copyright (C) 1997-2000, 2009-2012 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 @@ -20,25 +20,26 @@ #include #include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include - -#include "xalloc.h" +#include "data/casegrouper.h" +#include "data/casereader.h" +#include "data/casewriter.h" +#include "data/dataset.h" +#include "data/dictionary.h" +#include "data/transformations.h" +#include "data/variable.h" +#include "language/command.h" +#include "language/dictionary/split-file.h" +#include "language/lexer/lexer.h" +#include "language/lexer/variable-parser.h" +#include "libpspp/array.h" +#include "libpspp/assertion.h" +#include "libpspp/compiler.h" +#include "libpspp/i18n.h" +#include "libpspp/message.h" +#include "math/moments.h" +#include "output/tab.h" + +#include "gl/xalloc.h" #include "gettext.h" #define _(msgid) gettext (msgid) @@ -74,6 +75,9 @@ struct dsc_trns size_t var_cnt; /* Number of variables. */ enum dsc_missing_type missing_type; /* Treatment of missing values. */ enum mv_class exclude; /* Classes of missing values to exclude. */ + struct casereader *z_reader; /* Reader for count, mean, stddev. */ + casenumber count; /* Number left in this SPLIT FILE group.*/ + bool ok; }; /* Statistics. Used as bit indexes, so must be 32 or fewer. */ @@ -100,13 +104,13 @@ struct dsc_statistic_info static const struct dsc_statistic_info dsc_info[DSC_N_STATS] = { {"MEAN", N_("Mean"), MOMENT_MEAN}, - {"SEMEAN", N_("S E Mean"), MOMENT_VARIANCE}, + {"SEMEAN", N_("S.E. Mean"), MOMENT_VARIANCE}, {"STDDEV", N_("Std Dev"), MOMENT_VARIANCE}, {"VARIANCE", N_("Variance"), MOMENT_VARIANCE}, {"KURTOSIS", N_("Kurtosis"), MOMENT_KURTOSIS}, - {"SEKURTOSIS", N_("S E Kurt"), MOMENT_NONE}, + {"SEKURTOSIS", N_("S.E. Kurt"), MOMENT_NONE}, {"SKEWNESS", N_("Skewness"), MOMENT_SKEWNESS}, - {"SESKEWNESS", N_("S E Skew"), MOMENT_NONE}, + {"SESKEWNESS", N_("S.E. Skew"), MOMENT_NONE}, {"RANGE", N_("Range"), MOMENT_NONE}, {"MINIMUM", N_("Minimum"), MOMENT_NONE}, {"MAXIMUM", N_("Maximum"), MOMENT_NONE}, @@ -123,7 +127,7 @@ static const struct dsc_statistic_info dsc_info[DSC_N_STATS] = struct dsc_var { const struct variable *v; /* Variable to calculate on. */ - char z_name[VAR_NAME_LEN + 1]; /* Name for z-score variable. */ + char *z_name; /* Name for z-score variable. */ double valid, missing; /* Valid, missing counts. */ struct moments *moments; /* Moments. */ double min, max; /* Maximum and mimimum values. */ @@ -160,6 +164,9 @@ struct dsc_proc unsigned long show_stats; /* Statistics to display. */ unsigned long calc_stats; /* Statistics to calculate. */ enum moment max_moment; /* Highest moment needed for stats. */ + + /* Z scores. */ + struct casewriter *z_writer; /* Mean and stddev per SPLIT FILE group. */ }; /* Parsing. */ @@ -169,9 +176,9 @@ static void free_dsc_proc (struct dsc_proc *); /* Z-score functions. */ 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, int *z_cnt); +static char *generate_z_varname (const struct dictionary *dict, + struct dsc_proc *dsc, + const char *name, int *z_cnt); static void dump_z_table (struct dsc_proc *); static void setup_z_trns (struct dsc_proc *, struct dataset *); @@ -213,14 +220,15 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) dsc->sort_by_stat = DSC_NONE; dsc->sort_ascending = 1; dsc->show_stats = dsc->calc_stats = DEFAULT_STATS; + dsc->z_writer = NULL; /* Parse DESCRIPTIVES. */ - while (lex_token (lexer) != '.') + while (lex_token (lexer) != T_ENDCMD) { if (lex_match_id (lexer, "MISSING")) { - lex_match (lexer, '='); - while (lex_token (lexer) != '.' && lex_token (lexer) != '/') + lex_match (lexer, T_EQUALS); + while (lex_token (lexer) != T_ENDCMD && lex_token (lexer) != T_SLASH) { if (lex_match_id (lexer, "VARIABLE")) dsc->missing_type = DSC_VARIABLE; @@ -233,15 +241,15 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) lex_error (lexer, NULL); goto error; } - lex_match (lexer, ','); + lex_match (lexer, T_COMMA); } } else if (lex_match_id (lexer, "SAVE")) save_z_scores = 1; else if (lex_match_id (lexer, "FORMAT")) { - lex_match (lexer, '='); - while (lex_token (lexer) != '.' && lex_token (lexer) != '/') + lex_match (lexer, T_EQUALS); + while (lex_token (lexer) != T_ENDCMD && lex_token (lexer) != T_SLASH) { if (lex_match_id (lexer, "LABELS")) dsc->show_var_labels = 1; @@ -260,14 +268,14 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) lex_error (lexer, NULL); goto error; } - lex_match (lexer, ','); + lex_match (lexer, T_COMMA); } } else if (lex_match_id (lexer, "STATISTICS")) { - lex_match (lexer, '='); + lex_match (lexer, T_EQUALS); dsc->show_stats = 0; - while (lex_token (lexer) != '.' && lex_token (lexer) != '/') + while (lex_token (lexer) != T_ENDCMD && lex_token (lexer) != T_SLASH) { if (lex_match (lexer, T_ALL)) dsc->show_stats |= (1ul << DSC_N_STATS) - 1; @@ -275,14 +283,14 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) dsc->show_stats |= DEFAULT_STATS; else dsc->show_stats |= 1ul << (match_statistic (lexer)); - lex_match (lexer, ','); + lex_match (lexer, T_COMMA); } if (dsc->show_stats == 0) dsc->show_stats = DEFAULT_STATS; } else if (lex_match_id (lexer, "SORT")) { - lex_match (lexer, '='); + lex_match (lexer, T_EQUALS); if (lex_match_id (lexer, "NAME")) dsc->sort_by_stat = DSC_NAME; else @@ -291,7 +299,7 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) if (dsc->sort_by_stat == DSC_NONE ) dsc->sort_by_stat = DSC_MEAN; } - if (lex_match (lexer, '(')) + if (lex_match (lexer, T_LPAREN)) { if (lex_match_id (lexer, "A")) dsc->sort_ascending = 1; @@ -299,18 +307,18 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) dsc->sort_ascending = 0; else lex_error (lexer, NULL); - lex_force_match (lexer, ')'); + lex_force_match (lexer, T_RPAREN); } } else if (var_cnt == 0) { - if (lex_look_ahead (lexer) == '=') + if (lex_next_token (lexer, 1) == T_EQUALS) { lex_match_id (lexer, "VARIABLES"); - lex_match (lexer, '='); + lex_match (lexer, T_EQUALS); } - while (lex_token (lexer) != '.' && lex_token (lexer) != '/') + while (lex_token (lexer) != T_ENDCMD && lex_token (lexer) != T_SLASH) { int i; @@ -323,28 +331,29 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) { struct dsc_var *dv = &dsc->vars[i]; dv->v = vars[i]; - dv->z_name[0] = '\0'; + dv->z_name = NULL; dv->moments = NULL; } dsc->var_cnt = var_cnt; - if (lex_match (lexer, '(')) + if (lex_match (lexer, T_LPAREN)) { if (lex_token (lexer) != T_ID) { lex_error (lexer, NULL); goto error; } - if (try_name (dict, dsc, lex_tokid (lexer))) + if (try_name (dict, dsc, lex_tokcstr (lexer))) { - strcpy (dsc->vars[dsc->var_cnt - 1].z_name, lex_tokid (lexer)); + struct dsc_var *dsc_var = &dsc->vars[dsc->var_cnt - 1]; + dsc_var->z_name = xstrdup (lex_tokcstr (lexer)); z_cnt++; } else msg (SE, _("Z-score variable name %s would be" - " a duplicate variable name."), lex_tokid (lexer)); + " a duplicate variable name."), lex_tokcstr (lexer)); lex_get (lexer); - if (!lex_force_match (lexer, ')')) + if (!lex_force_match (lexer, T_RPAREN)) goto error; } } @@ -355,7 +364,7 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) goto error; } - lex_match (lexer, '/'); + lex_match (lexer, T_SLASH); } if (var_cnt == 0) { @@ -366,20 +375,34 @@ cmd_descriptives (struct lexer *lexer, struct dataset *ds) /* Construct z-score varnames, show translation table. */ if (z_cnt || save_z_scores) { + struct caseproto *proto; + if (save_z_scores) { int gen_cnt = 0; for (i = 0; i < dsc->var_cnt; i++) - if (dsc->vars[i].z_name[0] == 0) - { - if (!generate_z_varname (dict, dsc, dsc->vars[i].z_name, - var_get_name (dsc->vars[i].v), - &gen_cnt)) - goto error; - z_cnt++; - } + { + struct dsc_var *dsc_var = &dsc->vars[i]; + if (dsc_var->z_name == NULL) + { + const char *name = var_get_name (dsc_var->v); + dsc_var->z_name = generate_z_varname (dict, dsc, name, + &gen_cnt); + if (dsc_var->z_name == NULL) + goto error; + + z_cnt++; + } + } } + + proto = caseproto_create (); + for (i = 0; i < 1 + 2 * z_cnt; i++) + proto = caseproto_add_width (proto, 0); + dsc->z_writer = autopaging_writer_create (proto); + caseproto_unref (proto); + dump_z_table (dsc); } @@ -464,7 +487,12 @@ free_dsc_proc (struct dsc_proc *dsc) return; for (i = 0; i < dsc->var_cnt; i++) - moments_destroy (dsc->vars[i].moments); + { + struct dsc_var *dsc_var = &dsc->vars[i]; + free (dsc_var->z_name); + moments_destroy (dsc_var->moments); + } + casewriter_destroy (dsc->z_writer); free (dsc->vars); free (dsc); } @@ -482,33 +510,38 @@ try_name (const struct dictionary *dict, struct dsc_proc *dsc, 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 false; + { + struct dsc_var *dsc_var = &dsc->vars[i]; + if (dsc_var->z_name != NULL && !strcasecmp (dsc_var->z_name, name)) + 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 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, + known to already exist. If successful, returns the new name + as a dynamically allocated string. On failure, returns NULL. */ +static char * +generate_z_varname (const struct dictionary *dict, struct dsc_proc *dsc, const char *var_name, int *z_cnt) { - char name[VAR_NAME_LEN + 1]; + char *z_name, *trunc_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 (dict, dsc, name)) - { - strcpy (z_name, name); - return true; - } + z_name = xasprintf ("Z%s", var_name); + trunc_name = utf8_encoding_trunc (z_name, dict_get_encoding (dict), + ID_MAX_LEN); + free (z_name); + if (try_name (dict, dsc, trunc_name)) + return trunc_name; + free (trunc_name); /* Generate a synthetic name. */ for (;;) { + char name[8]; + (*z_cnt)++; if (*z_cnt <= 99) @@ -524,14 +557,11 @@ generate_z_varname (const struct dictionary *dict, struct dsc_proc *dsc, char *z 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 false; + return NULL; } if (try_name (dict, dsc, name)) - { - strcpy (z_name, name); - return true; - } + return xstrdup (name); } NOT_REACHED(); } @@ -548,27 +578,25 @@ dump_z_table (struct dsc_proc *dsc) size_t i; for (i = 0; i < dsc->var_cnt; i++) - if (dsc->vars[i].z_name[0] != '\0') + if (dsc->vars[i].z_name != NULL) cnt++; } - t = tab_create (2, cnt + 1, 0); + t = tab_create (2, cnt + 1); 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); tab_hline (t, TAL_2, 0, 1, 1); tab_text (t, 0, 0, TAB_CENTER | TAT_TITLE, _("Source")); tab_text (t, 1, 0, TAB_CENTER | TAT_TITLE, _("Target")); - tab_dim (t, tab_natural_dimensions, NULL, NULL); { size_t i, y; for (i = 0, y = 1; i < dsc->var_cnt; i++) - if (dsc->vars[i].z_name[0] != '\0') + if (dsc->vars[i].z_name != NULL) { - tab_text (t, 0, y, TAB_LEFT, var_get_name (dsc->vars[i].v)); + tab_text (t, 0, y, TAB_LEFT, var_to_string (dsc->vars[i].v)); tab_text (t, 1, y++, TAB_LEFT, dsc->vars[i].z_name); } } @@ -591,6 +619,36 @@ descriptives_trns_proc (void *trns_, struct ccase **c, const struct variable **vars; int all_sysmis = 0; + if (t->count <= 0) + { + struct ccase *z_case; + + z_case = casereader_read (t->z_reader); + if (z_case) + { + size_t z_idx = 0; + + t->count = case_num_idx (z_case, z_idx++); + for (z = t->z_scores; z < t->z_scores + t->z_score_cnt; z++) + { + z->mean = case_num_idx (z_case, z_idx++); + z->std_dev = case_num_idx (z_case, z_idx++); + } + case_unref (z_case); + } + else + { + if (t->ok) + { + msg (SE, _("Internal error processing Z scores")); + t->ok = false; + } + for (z = t->z_scores; z < t->z_scores + t->z_score_cnt; z++) + z->mean = z->std_dev = SYSMIS; + } + } + t->count--; + if (t->missing_type == DSC_LISTWISE) { assert(t->vars); @@ -625,11 +683,13 @@ static bool descriptives_trns_free (void *trns_) { struct dsc_trns *t = trns_; + bool ok = t->ok && !casereader_error (t->z_reader); free (t->z_scores); + casereader_destroy (t->z_reader); assert((t->missing_type != DSC_LISTWISE) ^ (t->vars != NULL)); free (t->vars); - return true; + return ok; } /* Sets up a transformation to calculate Z scores. */ @@ -640,7 +700,7 @@ setup_z_trns (struct dsc_proc *dsc, struct dataset *ds) size_t cnt, i; for (cnt = i = 0; i < dsc->var_cnt; i++) - if (dsc->vars[i].z_name[0] != '\0') + if (dsc->vars[i].z_name != NULL) cnt++; t = xmalloc (sizeof *t); @@ -660,24 +720,27 @@ setup_z_trns (struct dsc_proc *dsc, struct dataset *ds) t->var_cnt = 0; t->vars = NULL; } + t->z_reader = casewriter_make_reader (dsc->z_writer); + t->count = 0; + t->ok = true; + dsc->z_writer = NULL; for (cnt = i = 0; i < dsc->var_cnt; i++) { struct dsc_var *dv = &dsc->vars[i]; - if (dv->z_name[0] != '\0') + if (dv->z_name != NULL) { struct dsc_z_score *z; struct variable *dst_var; 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))); + var_set_label (dst_var, + xasprintf (_("Z-score of %s"),var_to_string (dv->v)), + false); z = &t->z_scores[cnt++]; z->src_var = dv->v; z->z_var = dst_var; - z->mean = dv->stats[DSC_MEAN]; - z->std_dev = dv->stats[DSC_STDDEV]; } } @@ -696,7 +759,9 @@ calc_descriptives (struct dsc_proc *dsc, struct casereader *group, struct dataset *ds) { struct casereader *pass1, *pass2; + casenumber count; struct ccase *c; + size_t z_idx; size_t i; c = casereader_peek (group, 0); @@ -728,6 +793,7 @@ calc_descriptives (struct dsc_proc *dsc, struct casereader *group, dsc->valid = 0.; /* First pass to handle most of the work. */ + count = 0; for (; (c = casereader_read (pass1)) != NULL; case_unref (c)) { double weight = dict_get_case_weight (dataset_dict (ds), c, NULL); @@ -760,6 +826,8 @@ calc_descriptives (struct dsc_proc *dsc, struct casereader *group, if (x > dv->max) dv->max = x; } + + count++; } if (!casereader_destroy (pass1)) { @@ -795,6 +863,15 @@ calc_descriptives (struct dsc_proc *dsc, struct casereader *group, } /* Calculate results. */ + if (dsc->z_writer) + { + c = case_create (casewriter_get_proto (dsc->z_writer)); + z_idx = 0; + case_data_rw_idx (c, z_idx++)->f = count; + } + else + c = NULL; + for (i = 0; i < dsc->var_cnt; i++) { struct dsc_var *dv = &dsc->vars[i]; @@ -828,8 +905,17 @@ calc_descriptives (struct dsc_proc *dsc, struct casereader *group, dv->stats[DSC_MAX] = dv->max == -DBL_MAX ? SYSMIS : dv->max; if (dsc->calc_stats & (1ul << DSC_SUM)) dv->stats[DSC_SUM] = W * dv->stats[DSC_MEAN]; + + if (dv->z_name) + { + case_data_rw_idx (c, z_idx++)->f = dv->stats[DSC_MEAN]; + case_data_rw_idx (c, z_idx++)->f = dv->stats[DSC_STDDEV]; + } } + if (c != NULL) + casewriter_write (dsc->z_writer, c); + /* Output results. */ display (dsc); } @@ -873,13 +959,12 @@ display (struct dsc_proc *dsc) sort (dsc->vars, dsc->var_cnt, sizeof *dsc->vars, descriptives_compare_dsc_vars, dsc); - t = tab_create (nc, dsc->var_cnt + 1, 0); + t = tab_create (nc, dsc->var_cnt + 1); tab_headers (t, 1, 0, 1, 0); tab_box (t, TAL_1, TAL_1, -1, -1, 0, 0, nc - 1, dsc->var_cnt); tab_box (t, -1, -1, -1, TAL_1, 1, 0, nc - 1, dsc->var_cnt); tab_hline (t, TAL_2, 0, nc - 1, 1); tab_vline (t, TAL_2, 1, 0, dsc->var_cnt); - tab_dim (t, tab_natural_dimensions, NULL, NULL); nc = 0; tab_text (t, nc++, 0, TAB_LEFT | TAT_TITLE, _("Variable")); @@ -904,10 +989,10 @@ display (struct dsc_proc *dsc) size_t j; nc = 0; - tab_text (t, nc++, i + 1, TAB_LEFT, var_get_name (dv->v)); - tab_text (t, nc++, i + 1, TAT_PRINTF, "%g", dv->valid); + tab_text (t, nc++, i + 1, TAB_LEFT, var_to_string (dv->v)); + tab_text_format (t, nc++, i + 1, 0, "%g", dv->valid); if (dsc->format == DSC_SERIAL) - tab_text (t, nc++, i + 1, TAT_PRINTF, "%g", dv->missing); + tab_text_format (t, nc++, i + 1, 0, "%g", dv->missing); for (j = 0; j < DSC_N_STATS; j++) if (dsc->show_stats & (1ul << j))