/* PSPP - a program for statistical analysis.
- Copyright (C) 2004, 2008, 2009 Free Software Foundation, Inc.
+ Copyright (C) 2004, 2008, 2009, 2010 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 <config.h>
#include <gsl/gsl_cdf.h>
-#include <libpspp/message.h>
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include <language/dictionary/split-file.h>
#include <language/lexer/lexer.h>
#include <libpspp/compiler.h>
-#include <libpspp/hash.h>
#include <libpspp/message.h>
#include <libpspp/misc.h>
#include <libpspp/str.h>
#include <math/moments.h>
-#include <output/charts/box-whisker.h>
-#include <output/charts/cartesian.h>
-#include <output/manager.h>
-#include <output/table.h>
+#include <output/chart-item.h>
+#include <output/charts/boxplot.h>
+#include <output/charts/np-plot.h>
+#include <output/tab.h>
#include "minmax.h"
#include "xalloc.h"
#define N_(msgid) msgid
/* (headers) */
-#include <output/chart.h>
#include <output/charts/plot-hist.h>
-#include <output/charts/plot-chart.h>
#include <math/histogram.h>
/* (specification)
struct percentile **ptl;
size_t n_ptiles;
- struct statistic *tukey_hinges;
- struct statistic *box_whisker;
- struct statistic *trimmed_mean;
- struct statistic *histogram;
- struct order_stats *np;
+ struct tukey_hinges *tukey_hinges;
+ struct box_whisker *box_whisker;
+ struct trimmed_mean *trimmed_mean;
+ struct histogram *histogram;
+ struct np *np;
/* Three quartiles indexing into PTL */
struct percentile **quartiles;
moments1_destroy (result->metrics[v].moments);
extrema_destroy (result->metrics[v].minima);
extrema_destroy (result->metrics[v].maxima);
- statistic_destroy (result->metrics[v].trimmed_mean);
- statistic_destroy (result->metrics[v].tukey_hinges);
- statistic_destroy (result->metrics[v].box_whisker);
- statistic_destroy (result->metrics[v].histogram);
+ statistic_destroy (&result->metrics[v].trimmed_mean->parent.parent);
+ statistic_destroy (&result->metrics[v].tukey_hinges->parent.parent);
+ statistic_destroy (&result->metrics[v].box_whisker->parent.parent);
+ statistic_destroy (&result->metrics[v].histogram->parent);
for (i = 0 ; i < result->metrics[v].n_ptiles; ++i)
- statistic_destroy ((struct statistic *) result->metrics[v].ptl[i]);
+ statistic_destroy (&result->metrics[v].ptl[i]->parent.parent);
free (result->metrics[v].ptl);
free (result->metrics[v].quartiles);
casereader_destroy (result->metrics[v].up_reader);
};
-/* Plot the normal and detrended normal plots for RESULT.
- Label the plots with LABEL */
-static void
-np_plot (struct np *np, const char *label)
-{
- double yfirst = 0, ylast = 0;
-
- double x_lower;
- double x_upper;
- double slack;
-
- /* Normal Plot */
- struct chart *np_chart;
-
- /* Detrended Normal Plot */
- struct chart *dnp_chart;
-
- /* The slope and intercept of the ideal normal probability line */
- const double slope = 1.0 / np->stddev;
- const double intercept = -np->mean / np->stddev;
-
- if ( np->n < 1.0 )
- {
- msg (MW, _("Not creating plot because data set is empty."));
- return ;
- }
-
- np_chart = chart_create ();
- dnp_chart = chart_create ();
-
- if ( !np_chart || ! dnp_chart )
- return ;
-
- chart_write_title (np_chart, _("Normal Q-Q Plot of %s"), label);
- chart_write_xlabel (np_chart, _("Observed Value"));
- chart_write_ylabel (np_chart, _("Expected Normal"));
-
- chart_write_title (dnp_chart, _("Detrended Normal Q-Q Plot of %s"),
- label);
- chart_write_xlabel (dnp_chart, _("Observed Value"));
- chart_write_ylabel (dnp_chart, _("Dev from Normal"));
-
- yfirst = gsl_cdf_ugaussian_Pinv (1 / (np->n + 1));
- ylast = gsl_cdf_ugaussian_Pinv (np->n / (np->n + 1));
-
- /* Need to make sure that both the scatter plot and the ideal fit into the
- plot */
- x_lower = MIN (np->y_min, (yfirst - intercept) / slope) ;
- x_upper = MAX (np->y_max, (ylast - intercept) / slope) ;
- slack = (x_upper - x_lower) * 0.05 ;
-
- chart_write_xscale (np_chart, x_lower - slack, x_upper + slack, 5);
- chart_write_xscale (dnp_chart, np->y_min, np->y_max, 5);
-
- chart_write_yscale (np_chart, yfirst, ylast, 5);
- chart_write_yscale (dnp_chart, np->dns_min, np->dns_max, 5);
-
- {
- struct casereader *reader = casewriter_make_reader (np->writer);
- struct ccase *c;
- while ((c = casereader_read (reader)) != NULL)
- {
- chart_datum (np_chart, 0, case_data_idx (c, NP_IDX_Y)->f, case_data_idx (c, NP_IDX_NS)->f);
- chart_datum (dnp_chart, 0, case_data_idx (c, NP_IDX_Y)->f, case_data_idx (c, NP_IDX_DNS)->f);
-
- case_unref (c);
- }
- casereader_destroy (reader);
- }
-
- chart_line (dnp_chart, 0, 0, np->y_min, np->y_max , CHART_DIM_X);
- chart_line (np_chart, slope, intercept, yfirst, ylast , CHART_DIM_Y);
-
- chart_submit (np_chart);
- chart_submit (dnp_chart);
-}
static void
ll != ll_null (&fctr->result_list);
ll = ll_next (ll))
{
- struct string str;
+ struct string label;
const struct factor_result *result =
ll_data (ll, struct factor_result, ll);
-
- ds_init_empty (&str);
- ds_put_format (&str, "%s ", var_get_name (dependent_var[v]));
-
- factor_to_string (fctr, result, &str);
-
- np_plot ((struct np*) result->metrics[v].np, ds_cstr(&str));
-
- statistic_destroy ((struct statistic *)result->metrics[v].np);
-
- ds_destroy (&str);
+ struct chart_item *npp, *dnpp;
+ struct casereader *reader;
+ struct np *np;
+
+ ds_init_empty (&label);
+ ds_put_format (&label, "%s ", var_get_name (dependent_var[v]));
+ factor_to_string (fctr, result, &label);
+
+ np = result->metrics[v].np;
+ reader = casewriter_make_reader (np->writer);
+ npp = np_plot_create (np, reader, ds_cstr (&label));
+ dnpp = dnp_plot_create (np, reader, ds_cstr (&label));
+
+ ds_destroy (&label);
+
+ if (npp == NULL || dnpp == NULL)
+ {
+ msg (MW, _("Not creating NP plot because data set is empty."));
+ chart_item_unref (npp);
+ chart_item_unref (dnpp);
+ }
+ else
+ {
+ chart_item_submit (npp);
+ chart_item_submit (dnpp);
+ }
+
+ statistic_destroy (&np->parent.parent);
+ casereader_destroy (reader);
}
}
}
struct string str;
const struct factor_result *result =
ll_data (ll, struct factor_result, ll);
+ struct histogram *histogram;
+ double mean, var, n;
+
+ histogram = result->metrics[v].histogram;
+ if (histogram == NULL)
+ {
+ /* Probably all values are SYSMIS. */
+ continue;
+ }
ds_init_empty (&str);
ds_put_format (&str, "%s ", var_get_name (dependent_var[v]));
factor_to_string (fctr, result, &str);
- histogram_plot ((struct histogram *) result->metrics[v].histogram,
- ds_cstr (&str),
- (struct moments1 *) result->metrics[v].moments);
+ moments1_calculate (result->metrics[v].moments,
+ &n, &mean, &var, NULL, NULL);
+ chart_item_submit (histogram_chart_create (histogram->gsl_hist,
+ ds_cstr (&str), n, mean,
+ sqrt (var), false));
ds_destroy (&str);
}
for (v = 0; v < n_dep_var; ++v)
{
- struct ll *ll;
- int f = 0;
- struct chart *ch = chart_create ();
+ const struct factor_result *result;
+ struct boxplot *boxplot;
double y_min = DBL_MAX;
double y_max = -DBL_MAX;
+ char *title;
- for (ll = ll_head (&fctr->result_list);
- ll != ll_null (&fctr->result_list);
- ll = ll_next (ll))
+ ll_for_each (result, struct factor_result, ll, &fctr->result_list)
{
+ struct factor_metrics *metrics = &result->metrics[v];
+ const struct ll_list *max_list = extrema_list (metrics->maxima);
+ const struct ll_list *min_list = extrema_list (metrics->minima);
const struct extremum *max, *min;
- const struct factor_result *result =
- ll_data (ll, struct factor_result, ll);
-
- const struct ll_list *max_list =
- extrema_list (result->metrics[v].maxima);
-
- const struct ll_list *min_list =
- extrema_list (result->metrics[v].minima);
if ( ll_is_empty (max_list))
{
continue;
}
- max = (const struct extremum *)
- ll_data (ll_head(max_list), struct extremum, ll);
-
- min = (const struct extremum *)
- ll_data (ll_head (min_list), struct extremum, ll);
+ max = ll_data (ll_head(max_list), struct extremum, ll);
+ min = ll_data (ll_head (min_list), struct extremum, ll);
y_max = MAX (y_max, max->value);
y_min = MIN (y_min, min->value);
}
- boxplot_draw_yscale (ch, y_max, y_min);
-
- if ( fctr->indep_var[0])
- chart_write_title (ch, _("Boxplot of %s vs. %s"),
+ if (fctr->indep_var[0])
+ title = xasprintf (_("Boxplot of %s vs. %s"),
var_to_string (dependent_var[v]),
- var_to_string (fctr->indep_var[0]) );
+ var_to_string (fctr->indep_var[0]));
else
- chart_write_title (ch, _("Boxplot of %s"),
- var_to_string (dependent_var[v]));
+ title = xasprintf (_("Boxplot of %s"),
+ var_to_string (dependent_var[v]));
+ boxplot = boxplot_create (y_min, y_max, title);
+ free (title);
- for (ll = ll_head (&fctr->result_list);
- ll != ll_null (&fctr->result_list);
- ll = ll_next (ll))
+ ll_for_each (result, struct factor_result, ll, &fctr->result_list)
{
- const struct factor_result *result =
- ll_data (ll, struct factor_result, ll);
-
- struct string str;
- const double box_width = (ch->data_right - ch->data_left)
- / (ll_count (&fctr->result_list) * 2.0 ) ;
-
- const double box_centre = (f++ * 2 + 1) * box_width + ch->data_left;
-
- ds_init_empty (&str);
+ struct factor_metrics *metrics = &result->metrics[v];
+ struct string str = DS_EMPTY_INITIALIZER;
factor_to_string_concise (fctr, result, &str);
-
- boxplot_draw_boxplot (ch,
- box_centre, box_width,
- (const struct box_whisker *)
- result->metrics[v].box_whisker,
- ds_cstr (&str));
-
+ boxplot_add_box (boxplot, metrics->box_whisker, ds_cstr (&str));
+ metrics->box_whisker = NULL;
ds_destroy (&str);
}
- chart_submit (ch);
+ boxplot_submit (boxplot);
}
}
)
{
+ const struct factor_result *result;
int v;
- struct ll *ll;
- const struct ll_list *result_list = &fctr->result_list;
-
- for (ll = ll_head (result_list);
- ll != ll_null (result_list);
- ll = ll_next (ll))
+ ll_for_each (result, struct factor_result, ll, &fctr->result_list)
{
struct string title;
- struct chart *ch = chart_create ();
double y_min = DBL_MAX;
double y_max = -DBL_MAX;
-
- const struct factor_result *result =
- ll_data (ll, struct factor_result, ll);
-
- const double box_width = (ch->data_right - ch->data_left)
- / (n_dep_var * 2.0 ) ;
+ struct boxplot *boxplot;
for (v = 0; v < n_dep_var; ++v)
{
- const struct ll *max_ll =
- ll_head (extrema_list (result->metrics[v].maxima));
- const struct ll *min_ll =
- ll_head (extrema_list (result->metrics[v].minima));
-
- const struct extremum *max =
- (const struct extremum *) ll_data (max_ll, struct extremum, ll);
-
- const struct extremum *min =
- (const struct extremum *) ll_data (min_ll, struct extremum, ll);
+ const struct factor_metrics *metrics = &result->metrics[v];
+ const struct ll *max_ll = ll_head (extrema_list (metrics->maxima));
+ const struct ll *min_ll = ll_head (extrema_list (metrics->minima));
+ const struct extremum *max = ll_data (max_ll, struct extremum, ll);
+ const struct extremum *min = ll_data (min_ll, struct extremum, ll);
y_max = MAX (y_max, max->value);
y_min = MIN (y_min, min->value);
}
-
- boxplot_draw_yscale (ch, y_max, y_min);
-
ds_init_empty (&title);
factor_to_string (fctr, result, &title);
-
-#if 0
- ds_put_format (&title, "%s = ", var_get_name (fctr->indep_var[0]));
- var_append_value_name (fctr->indep_var[0], &result->value[0], &title);
-#endif
-
- chart_write_title (ch, ds_cstr (&title));
+ boxplot = boxplot_create (y_min, y_max, ds_cstr (&title));
ds_destroy (&title);
for (v = 0; v < n_dep_var; ++v)
{
- struct string str;
- const double box_centre = (v * 2 + 1) * box_width + ch->data_left;
-
- ds_init_empty (&str);
- ds_init_cstr (&str, var_get_name (dependent_var[v]));
-
- boxplot_draw_boxplot (ch,
- box_centre, box_width,
- (const struct box_whisker *) result->metrics[v].box_whisker,
- ds_cstr (&str));
-
- ds_destroy (&str);
+ struct factor_metrics *metrics = &result->metrics[v];
+ boxplot_add_box (boxplot, metrics->box_whisker,
+ var_get_name (dependent_var[v]));
+ metrics->box_whisker = NULL;
}
- chart_submit (ch);
+ boxplot_submit (boxplot);
}
}
if ( cmd.sbc_percentiles)
show_percentiles (dependent_vars, n_dependent_vars, factor);
- if (cmd.a_plot[XMN_PLT_BOXPLOT] &&
- cmd.cmp == XMN_GROUPS)
- show_boxplot_groups (dependent_vars, n_dependent_vars, factor);
-
-
- if (cmd.a_plot[XMN_PLT_BOXPLOT] &&
- cmd.cmp == XMN_VARIABLES)
- show_boxplot_variables (dependent_vars, n_dependent_vars,
- factor);
-
+ if (cmd.a_plot[XMN_PLT_BOXPLOT])
+ {
+ if (cmd.cmp == XMN_GROUPS)
+ show_boxplot_groups (dependent_vars, n_dependent_vars, factor);
+ else if (cmd.cmp == XMN_VARIABLES)
+ show_boxplot_variables (dependent_vars, n_dependent_vars, factor);
+ }
+
if (cmd.a_plot[XMN_PLT_HISTOGRAM])
show_histogram (dependent_vars, n_dependent_vars, factor);
xmn_custom_percentiles (struct lexer *lexer, struct dataset *ds UNUSED,
struct cmd_examine *p UNUSED, void *aux UNUSED)
{
- lex_match (lexer, '=');
+ lex_match (lexer, T_EQUALS);
- lex_match (lexer, '(');
+ lex_match (lexer, T_LPAREN);
while ( lex_is_number (lexer) )
{
lex_get (lexer);
- lex_match (lexer, ',') ;
+ lex_match (lexer, T_COMMA) ;
}
- lex_match (lexer, ')');
+ lex_match (lexer, T_RPAREN);
- lex_match (lexer, '=');
+ lex_match (lexer, T_EQUALS);
if ( lex_match_id (lexer, "HAVERAGE"))
percentile_algorithm = PC_HAVERAGE;
void *aux UNUSED)
{
const struct dictionary *dict = dataset_dict (ds);
- lex_match (lexer, '=');
+ lex_match (lexer, T_EQUALS);
- if ( (lex_token (lexer) != T_ID || dict_lookup_var (dict, lex_tokid (lexer)) == NULL)
+ if ( (lex_token (lexer) != T_ID || dict_lookup_var (dict, lex_tokcstr (lexer)) == NULL)
&& lex_token (lexer) != T_ALL)
{
return 2;
ll_init (&sf->result_list);
if ( (lex_token (lexer) != T_ID ||
- dict_lookup_var (dict, lex_tokid (lexer)) == NULL)
+ dict_lookup_var (dict, lex_tokcstr (lexer)) == NULL)
&& lex_token (lexer) != T_ALL)
{
free ( sf ) ;
lex_match (lexer, T_BY);
if ( (lex_token (lexer) != T_ID ||
- dict_lookup_var (dict, lex_tokid (lexer)) == NULL)
+ dict_lookup_var (dict, lex_tokcstr (lexer)) == NULL)
&& lex_token (lexer) != T_ALL)
{
free (sf);
else
ll_push_tail (&factor_list, &sf->ll);
- lex_match (lexer, ',');
+ lex_match (lexer, T_COMMA);
- if ( lex_token (lexer) == '.' || lex_token (lexer) == '/' )
+ if ( lex_token (lexer) == T_ENDCMD || lex_token (lexer) == T_SLASH )
return 1;
success = examine_parse_independent_vars (lexer, dict, cmd);
metric->n_ptiles = percentile_list.n_data;
- metric->ptl = xcalloc (metric->n_ptiles,
- sizeof (struct percentile *));
+ metric->ptl = xcalloc (metric->n_ptiles, sizeof *metric->ptl);
metric->quartiles = xcalloc (3, sizeof (*metric->quartiles));
for (i = 0 ; i < metric->n_ptiles; ++i)
{
- metric->ptl[i] = (struct percentile *)
- percentile_create (percentile_list.data[i] / 100.0, metric->n_valid);
+ metric->ptl[i] = percentile_create (percentile_list.data[i] / 100.0, metric->n_valid);
if ( percentile_list.data[i] == 25)
metric->quartiles[0] = metric->ptl[i];
n_os ++;
}
- os = xcalloc (sizeof (struct order_stats *), n_os);
+ os = xcalloc (n_os, sizeof *os);
for (i = 0 ; i < metric->n_ptiles ; ++i )
{
- os[i] = (struct order_stats *) metric->ptl[i];
+ os[i] = &metric->ptl[i]->parent;
}
- os[i] = (struct order_stats *) metric->tukey_hinges;
- os[i+1] = (struct order_stats *) metric->trimmed_mean;
+ os[i] = &metric->tukey_hinges->parent;
+ os[i+1] = &metric->trimmed_mean->parent;
if (cmd->a_plot[XMN_PLT_NPPLOT])
- os[i+2] = metric->np;
+ os[i+2] = &metric->np->parent;
order_stats_accumulate (os, n_os,
casereader_clone (metric->up_reader),
{
struct factor_metrics *metric = &result->metrics[v];
if ( metric->histogram)
- histogram_add ((struct histogram *) metric->histogram,
+ histogram_add (metric->histogram,
case_data (c, dependent_vars[v])->f, weight);
}
case_unref (c);
struct factor_metrics *metric = &result->metrics[v];
int n_vals = caseproto_get_n_widths (casereader_get_proto (
metric->up_reader));
+ struct order_stats *os;
metric->box_whisker =
- box_whisker_create ((struct tukey_hinges *) metric->tukey_hinges,
- cmd->v_id, n_vals - 1);
+ box_whisker_create ( metric->tukey_hinges, cmd->v_id, n_vals - 1);
- order_stats_accumulate ((struct order_stats **) &metric->box_whisker,
- 1,
+ os = &metric->box_whisker->parent;
+ order_stats_accumulate ( &os, 1,
casereader_clone (metric->up_reader),
wv, dependent_vars[v], MV_ANY);
}
n_cols = heading_columns + 6;
- tbl = tab_create (n_cols, n_rows, 0);
+ tbl = tab_create (n_cols, n_rows);
tab_headers (tbl, heading_columns, 0, heading_rows, 0);
- tab_dim (tbl, tab_natural_dimensions, NULL);
-
/* Outline the box */
tab_box (tbl,
TAL_2, TAL_2,
n_cols = heading_columns + 2;
- tbl = tab_create (n_cols, n_rows, 0);
+ tbl = tab_create (n_cols, n_rows);
tab_headers (tbl, heading_columns, 0, heading_rows, 0);
- tab_dim (tbl, tab_natural_dimensions, NULL);
-
/* Outline the box */
tab_box (tbl,
TAL_2, TAL_2,
tab_double (tbl, n_cols - 2,
heading_rows + row_var_start + 3 + i * DESCRIPTIVE_ROWS,
TAB_CENTER,
- trimmed_mean_calculate ((struct trimmed_mean *) result->metrics[v].trimmed_mean),
+ trimmed_mean_calculate (result->metrics[v].trimmed_mean),
NULL);
n_cols = heading_columns + 2;
- tbl = tab_create (n_cols, n_rows, 0);
+ tbl = tab_create (n_cols, n_rows);
tab_headers (tbl, heading_columns, 0, heading_rows, 0);
- tab_dim (tbl, tab_natural_dimensions, NULL);
-
/* Outline the box */
tab_box (tbl,
TAL_2, TAL_2,
min_ll = ll_next (min_ll);
}
-
max_ll = ll_head (extrema_list (result->metrics[v].maxima));
for (e = 0; e < cmd.st_n;)
{
n_cols = heading_columns + n_percentiles;
- tbl = tab_create (n_cols, n_rows, 0);
+ tbl = tab_create (n_cols, n_rows);
tab_headers (tbl, heading_columns, 0, heading_rows, 0);
- tab_dim (tbl, tab_natural_dimensions, NULL);
-
/* Outline the box */
tab_box (tbl,
TAL_2, TAL_2,
tab_vline (tbl, TAL_1, n_cols - n_percentiles -1, heading_rows, n_rows - 1);
- tukey_hinges_calculate ((struct tukey_hinges *) result->metrics[v].tukey_hinges,
- hinges);
+ tukey_hinges_calculate (result->metrics[v].tukey_hinges, hinges);
for (j = 0; j < n_percentiles; ++j)
{