#include "output/chart-item.h"
#include "output/charts/piechart.h"
#include "output/charts/plot-hist.h"
+#include "output/pivot-table.h"
#include "output/tab.h"
#include "gl/minmax.h"
struct freq_tab
{
struct casewriter *sorter;
+ struct casereader *data;
struct freq *valid; /* Valid freqs. */
int n_valid; /* Number of valid freqs. */
/* Extract data from hash table. */
reader = casewriter_make_reader (ft->sorter);
+ ft->data = casereader_clone (reader);
freqs = xnmalloc (casereader_count_cases (reader), sizeof *freqs);
for (count = 0; (c = casereader_read (reader)) != NULL; count++)
{
for (i = 0; i < vf->tab.n_valid + vf->tab.n_missing; i++)
value_destroy (&vf->tab.valid[i].value, vf->width);
}
+ casereader_destroy (vf->tab.data);
free (vf->tab.valid);
}
static void
dump_freq_table (const struct var_freqs *vf, const struct variable *wv)
{
- const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : &F_8_0;
- const struct freq_tab *ft = &vf->tab;
- int n_categories;
- struct freq *f;
- struct tab_table *t;
- int r, x;
- double cum_total = 0.0;
- double cum_freq = 0.0;
-
- static const char *headings[] = {
- N_("Value Label"),
- N_("Value"),
- N_("Frequency"),
- N_("Percent"),
- N_("Valid Percent"),
- N_("Cum Percent")
- };
-
- n_categories = ft->n_valid + ft->n_missing;
- t = tab_create (6, n_categories + 2);
- tab_headers (t, 0, 0, 1, 0);
-
- for (x = 0; x < 6; x++)
- tab_text (t, x, 0, TAB_CENTER | TAT_TITLE, gettext (headings[x]));
-
- r = 1;
- for (f = ft->valid; f < ft->missing; f++)
- {
- const char *label;
- double percent, valid_percent;
+ struct variable *value_var, *freq_var;
+ struct pivot_table *pt;
- cum_freq += f->count;
+ pt = xmalloc (sizeof *pt);
- percent = f->count / ft->total_cases * 100.0;
- valid_percent = f->count / ft->valid_cases * 100.0;
- cum_total += valid_percent;
+ pt->data = casereader_clone (vf->tab.data);
- label = var_lookup_value_label (vf->var, &f->value);
- if (label != NULL)
- tab_text (t, 0, r, TAB_LEFT, label);
-
- tab_value (t, 1, r, TAB_NONE, &f->value, ft->dict, &vf->print);
- tab_double (t, 2, r, TAB_NONE, f->count, wfmt);
- tab_double (t, 3, r, TAB_NONE, percent, NULL);
- tab_double (t, 4, r, TAB_NONE, valid_percent, NULL);
- tab_double (t, 5, r, TAB_NONE, cum_total, NULL);
- r++;
- }
- for (; f < &ft->valid[n_categories]; f++)
+ pt->dict = dict_create ();
+ value_var = dict_clone_var_assert (pt->dict, vf->var);
+ if (wv != NULL)
+ freq_var = dict_clone_var_assert (pt->dict, wv);
+ else
{
- const char *label;
-
- cum_freq += f->count;
-
- label = var_lookup_value_label (vf->var, &f->value);
- if (label != NULL)
- tab_text (t, 0, r, TAB_LEFT, label);
-
- tab_value (t, 1, r, TAB_NONE, &f->value, ft->dict, &vf->print);
- tab_double (t, 2, r, TAB_NONE, f->count, wfmt);
- tab_double (t, 3, r, TAB_NONE,
- f->count / ft->total_cases * 100.0, NULL);
- tab_text (t, 4, r, TAB_NONE, _("Missing"));
- r++;
+ freq_var = dict_create_var (pt->dict, "$Frequency", 0);
+ var_set_both_formats (freq_var, &F_8_0);
+ var_set_label (freq_var, _("Frequency"));
}
- tab_box (t, TAL_1, TAL_1, -1, TAL_1, 0, 0, 5, r);
- tab_hline (t, TAL_2, 0, 5, 1);
- tab_hline (t, TAL_2, 0, 5, r);
- tab_joint_text (t, 0, r, 1, r, TAB_RIGHT | TAT_TITLE, _("Total"));
- tab_vline (t, TAL_0, 1, r, r);
- tab_double (t, 2, r, TAB_NONE, cum_freq, wfmt);
- tab_fixed (t, 3, r, TAB_NONE, 100.0, 5, 1);
- tab_fixed (t, 4, r, TAB_NONE, 100.0, 5, 1);
-
- tab_title (t, "%s", var_to_string (vf->var));
- tab_submit (t);
+ subcase_init_empty (&pt->split);
+ subcase_init_empty (&pt->dimensions[TABLE_HORZ]);
+ subcase_init_empty (&pt->dimensions[TABLE_VERT]);
+ subcase_add_var (&pt->dimensions[TABLE_VERT], value_var, SC_ASCEND);
+
+ pt->pane.n[TABLE_HORZ] = 4;
+ pt->pane.n[TABLE_VERT] = 1;
+ pt->pane.cells = xmalloc (sizeof *pt->pane.cells);
+ pt->pane.cells[0] = xmalloc (4 * sizeof **pt->pane.cells);
+
+ /* Frequency. */
+ pivot_cell_init (
+ &pt->pane.cells[0][0],
+ pivot_value_create (freq_var, PIVOT_SUM, 0, 0, PIVOT_INCLUDE_ALL));
+
+ /* Percent. */
+ pivot_cell_init (
+ &pt->pane.cells[0][1],
+ pivot_value_create (freq_var, PIVOT_SUM, 0, 0, PIVOT_INCLUDE_ALL));
+ pt->pane.cells[0][1].cmp = PIVOT_PERCENT;
+ pt->pane.cells[0][1].cmp_args[0] = pivot_value_create (
+ freq_var, PIVOT_SUM, 0, 1, PIVOT_INCLUDE_ALL);
+ pt->pane.cells[0][1].label = xstrdup ("Percent");
+
+ /* Valid Percent. */
+ pivot_cell_init (
+ &pt->pane.cells[0][2],
+ pivot_value_create (freq_var, PIVOT_SUM, 0, 0, PIVOT_INCLUDE_VALID));
+ pt->pane.cells[0][2].cmp = PIVOT_PERCENT;
+ pt->pane.cells[0][2].cmp_args[0] = pivot_value_create (
+ freq_var, PIVOT_SUM, 0, 1, PIVOT_INCLUDE_VALID);
+ pt->pane.cells[0][2].label = xstrdup ("Valid Percent");
+
+ /* Cumulative (Valid) Percent. */
+ pivot_cell_init (
+ &pt->pane.cells[0][3],
+ pivot_value_create (freq_var, PIVOT_SUM, 0, 0, PIVOT_INCLUDE_VALID));
+ pt->pane.cells[0][3].base->n_cum_vars[TABLE_VERT] = 1;
+ pt->pane.cells[0][3].cmp = PIVOT_PERCENT;
+ pt->pane.cells[0][3].cmp_args[0] = pivot_value_create (
+ freq_var, PIVOT_SUM, 0, 1, PIVOT_INCLUDE_VALID);
+ pt->pane.cells[0][3].label = xstrdup ("Cum Percent");
+
+ pivot_table_dump (pt);
}
\f
/* Statistical display. */
src/output/output-item-provider.h \
src/output/output-item.c \
src/output/output-item.h \
+ src/output/pivot-table.c \
+ src/output/pivot-table.h \
src/output/render.c \
src/output/render.h \
src/output/tab.c \
--- /dev/null
+/* PSPP - a program for statistical analysis.
+ Copyright (C) 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
+ the Free Software Foundation, either version 3 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program. If not, see <http://www.gnu.org/licenses/>. */
+
+#include <config.h>
+
+#include "output/pivot-table.h"
+
+#include "data/case.h"
+#include "data/casereader.h"
+#include "math/sort.h"
+
+#include "gl/xalloc.h"
+
+void
+pivot_cell_init (struct pivot_cell *cell, struct pivot_value *value)
+{
+ cell->base = value;
+ cell->cmp = PIVOT_RAW;
+ cell->cmp_args[0] = cell->cmp_args[1] = cell->cmp_args[2] = NULL;
+ cell->label = NULL;
+ cell->format = F_8_0;
+}
+
+struct pivot_value *
+pivot_value_create (const struct variable *var,
+ enum pivot_operator operator,
+ int n_vars_horz,
+ int n_vars_vert,
+ enum pivot_value_include include)
+{
+ struct pivot_value *value = xmalloc (sizeof *value);
+ value->var = var;
+ value->operator = operator;
+ value->n_agg_vars[TABLE_HORZ] = n_vars_horz;
+ value->n_agg_vars[TABLE_VERT] = n_vars_vert;
+ value->include = include;
+ return value;
+}
+
+static struct ccase *
+keep_one (struct ccase *a, struct ccase *b, void *aux UNUSED)
+{
+ case_unref (b);
+ return a;
+}
+
+static void
+dump_projection (struct casereader *data, const struct subcase *split,
+ const struct subcase *vars)
+{
+ struct subcase sc;
+ struct ccase *c;
+
+ subcase_clone (&sc, split);
+ subcase_concat_always (&sc, vars);
+
+ data = casereader_project (data, &sc);
+ data = sort_distinct_execute (data, &sc, keep_one, NULL, NULL);
+
+#if 0
+ for (; (c = casereader_read (data)) != NULL; case_unref (c))
+ {
+ const struct caseproto *proto = case_get_proto (c);
+ size_t i;
+
+ for (i = 0; i < caseproto_get_n_widths (proto); i++)
+ {
+ int width = caseproto_get_width (proto, i);
+
+ if (i > 0)
+ putchar (' ');
+
+ if (width == 0)
+ printf ("%8.2g", case_num_idx (c, i));
+ else
+ printf ("\"%.*s\"", width, case_str_idx (c, i));
+ }
+ printf ("\n");
+ }
+#endif
+
+ subcase_destroy (&sc);
+}
+
+static void
+dump_data (struct pivot_table *pt)
+{
+ struct casereader *data;
+ struct subcase sc;
+ struct ccase *c;
+
+ subcase_clone (&sc, &pt->split);
+ subcase_concat_always (&sc, &pt->dimensions[TABLE_HORZ]);
+ subcase_concat_always (&sc, &pt->dimensions[TABLE_VERT]);
+
+ data = sort_distinct_execute (casereader_clone (pt->data), &sc, keep_one, NULL, NULL);
+ for (; (c = casereader_read (data)) != NULL; case_unref (c))
+ {
+ const struct caseproto *proto = case_get_proto (c);
+ size_t i;
+
+ for (i = 0; i < caseproto_get_n_widths (proto); i++)
+ {
+ int width = caseproto_get_width (proto, i);
+
+ if (i > 0)
+ putchar (' ');
+
+ if (width == 0)
+ printf ("%8.2f", case_num_idx (c, i));
+ else
+ printf ("\"%.*s\"", width, case_str_idx (c, i));
+ }
+ printf ("\n");
+ }
+
+}
+
+void
+pivot_table_dump (struct pivot_table *pt)
+{
+ /* Strategy:
+
+ Determine row and column labels:
+
+ Project onto (splits, rows) and sort, discarding duplicates.
+
+ Project onto (splits, columns) and sort, discarding duplicates.
+
+ Sort data on (splits, rows, columns).
+
+ For each split group:
+
+ For each row value:
+
+ For each column value:
+
+ For each value in the pane:
+
+ Evaluate and print value.
+
+ */
+
+ dump_projection (casereader_clone (pt->data),
+ &pt->split, &pt->dimensions[TABLE_HORZ]);
+ dump_projection (casereader_clone (pt->data),
+ &pt->split, &pt->dimensions[TABLE_VERT]);
+
+ dump_data (pt);
+
+
+ casereader_destroy (pt->data);
+}
--- /dev/null
+/* PSPP - a program for statistical analysis.
+ Copyright (C) 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
+ the Free Software Foundation, either version 3 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with this program. If not, see <http://www.gnu.org/licenses/>. */
+
+#ifndef OUTPUT_PIVOT_TABLE_H
+#define OUTPUT_PIVOT_TABLE_H 1
+
+#include "data/format.h"
+#include "data/subcase.h"
+#include "output/table.h"
+
+struct pivot_pane
+ {
+ int n[TABLE_N_AXES];
+ struct pivot_cell **cells; /* cells[y][x] */
+ };
+
+struct pivot_table
+ {
+ struct casereader *data;
+ struct dictionary *dict;
+ struct subcase split;
+ struct subcase dimensions[TABLE_N_AXES];
+ struct pivot_pane pane;
+ struct pivot_pane *summaries[TABLE_N_AXES];
+ };
+
+void pivot_table_dump (struct pivot_table *);
+
+enum pivot_comparison_type
+ {
+ PIVOT_RAW, /* No comparison. */
+
+ /* 1-dimensional comparisons. */
+ PIVOT_FRACTION, /* Fraction of an aggregate. */
+ PIVOT_PERCENT, /* Percentage of an aggregate. */
+
+ /* 2-dimensional comparisons. */
+ PIVOT_EXPECTED, /* row_sum * col_sum / grand_total */
+ PIVOT_RESIDUAL, /* value - expected */
+ PIVOT_SRESIDUAL, /* (value - expected) / sqrt(expected) */
+ PIVOT_ASRESIDUAL /* (value - expected)
+ / sqrt(expected
+ * (1 - row_sum / grand_total)
+ * (1 - col_sum / grand_total)) */
+ };
+
+/* A value, possibly compared against other values. */
+struct pivot_cell
+ {
+ struct pivot_value *base;
+
+ /* Comparison. */
+ enum pivot_comparison_type cmp;
+ struct pivot_value *cmp_args[3];
+
+ /* Formatting. */
+ char *label;
+ struct fmt_spec format;
+ };
+
+void pivot_cell_init (struct pivot_cell *, struct pivot_value *);
+
+/* How to combine potentially multiple values. */
+enum pivot_operator
+ {
+ PIVOT_EXACTLY_ONE, /* SYSMIS if no values or more than one. */
+ PIVOT_SUM,
+ PIVOT_MEAN,
+
+ PIVOT_N,
+ PIVOT_N_VALID,
+ PIVOT_N_MISSING,
+
+ PIVOT_MODE,
+
+ PIVOT_STDDEV,
+ PIVOT_VARIANCE,
+
+ PIVOT_MINIMUM,
+ PIVOT_MAXIMUM,
+ PIVOT_RANGE,
+ PIVOT_MEDIAN,
+ PIVOT_PERCENTILE
+ };
+
+enum pivot_value_include
+ {
+ PIVOT_INCLUDE_VALID = 1 << 0,
+ PIVOT_INCLUDE_SYSTEM_MISSING = 1 << 1,
+ PIVOT_INCLUDE_USER_MISSING = 1 << 2,
+ PIVOT_INCLUDE_ALL = 7
+ };
+
+/* A pivot_value specifies a function whose arguments are the values of:
+
+ - The split and dimension variables of the pivot table.
+
+ - The variable specified as part of the pivot_value. (PIVOT_N does not
+ require a variable.)
+
+ A pivot_value is notionally evaluated as:
+
+ Given a value for every variable:
+ For every row in the input data:
+ If the row's split variable equal their assigned values:
+ and the row's dimension variables equal their assigned values,
+ except that the aggregate variables, if any, are ignored,
+ and that the cumulative variables are equal or lexicographically
+ less than the assigned values,
+ and the value variable is included according to "include":
+ Submit the value to the operator.
+*/
+struct pivot_value
+ {
+ const struct variable *var;
+ enum pivot_operator operator;
+ int n_agg_vars[TABLE_N_AXES];
+ int n_cum_vars[TABLE_N_AXES];
+ enum pivot_value_include include;
+ };
+
+struct pivot_value *pivot_value_create (const struct variable *,
+ enum pivot_operator,
+ int n_vars_horz,
+ int n_vars_vert,
+ enum pivot_value_include);
+
+
+#endif /* output/pivot-table.h */
projects / pspp / commitdiff