const struct ctables_cell *example;
- double valid;
- double missing;
+ double d_valid; /* Dictionary weight. */
+ double d_missing;
+ double e_valid; /* Effective weight */
+ double e_missing;
};
enum ctables_summary_variant
bool mrsets_count_duplicates; /* MRSETS. */
bool smissing_listwise; /* SMISSING. */
- struct variable *base_weight; /* WEIGHT. */
+ struct variable *e_weight; /* WEIGHT. */
int hide_threshold; /* HIDESMALLCOUNTS. */
struct ctables_table **tables;
ctables_summary_add (union ctables_summary *s,
const struct ctables_summary_spec *ss,
const struct variable *var, const union value *value,
- double weight)
+ double d_weight, double e_weight)
{
switch (ss->function)
{
case CTSF_COUNT:
+ case CSTF_TOTALN:
+ case CTSF_VALIDN:
+ if (var_is_value_missing (var, value))
+ s->missing += d_weight;
+ else
+ s->valid += d_weight;
+ break;
+
case CTSF_ECOUNT:
case CTSF_ROWPCT_COUNT:
case CTSF_COLPCT_COUNT:
case CTSF_LAYERROWPCT_TOTALN:
case CTSF_LAYERCOLPCT_TOTALN:
case CTSF_MISSING:
- case CSTF_TOTALN:
case CTSF_ETOTALN:
- case CTSF_VALIDN:
case CTSF_EVALIDN:
if (var_is_value_missing (var, value))
- s->missing += weight;
+ s->missing += e_weight;
else
- s->valid += weight;
+ s->valid += e_weight;
break;
case CTSF_MAXIMUM:
case CTSF_LAYERROWPCT_SUM:
case CTSF_LAYERCOLPCT_SUM:
if (!var_is_value_missing (var, value))
- moments1_add (s->moments, value->f, weight);
+ moments1_add (s->moments, value->f, e_weight);
break;
case CTSF_MEDIAN:
case CTSF_PTILE:
if (var_is_value_missing (var, value))
{
- s->ovalid += weight;
+ s->ovalid += e_weight;
struct ccase *c = case_create (casewriter_get_proto (s->writer));
*case_num_rw_idx (c, 0) = value->f;
- *case_num_rw_idx (c, 1) = weight;
+ *case_num_rw_idx (c, 1) = e_weight;
casewriter_write (s->writer, c);
}
break;
}
}
+static enum ctables_domain_type
+ctables_function_domain (enum ctables_summary_function function)
+{
+ switch (function)
+ {
+ case CTSF_COUNT:
+ case CTSF_ECOUNT:
+ case CTSF_MISSING:
+ case CSTF_TOTALN:
+ case CTSF_ETOTALN:
+ case CTSF_VALIDN:
+ case CTSF_EVALIDN:
+ case CTSF_MAXIMUM:
+ case CTSF_MINIMUM:
+ case CTSF_RANGE:
+ case CTSF_MEAN:
+ case CTSF_SEMEAN:
+ case CTSF_STDDEV:
+ case CTSF_SUM:
+ case CTSF_VARIANCE:
+ case CTSF_MEDIAN:
+ case CTSF_PTILE:
+ case CTSF_MODE:
+ case CTSF_RESPONSES:
+ NOT_REACHED ();
+
+ case CTSF_COLPCT_COUNT:
+ case CTSF_COLPCT_COUNT_RESPONSES:
+ case CTSF_COLPCT_RESPONSES:
+ case CTSF_COLPCT_RESPONSES_COUNT:
+ case CTSF_COLPCT_SUM:
+ case CTSF_COLPCT_TOTALN:
+ case CTSF_COLPCT_VALIDN:
+ return CTDT_COL;
+
+ case CTSF_LAYERCOLPCT_COUNT:
+ case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
+ case CTSF_LAYERCOLPCT_RESPONSES:
+ case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
+ case CTSF_LAYERCOLPCT_SUM:
+ case CTSF_LAYERCOLPCT_TOTALN:
+ case CTSF_LAYERCOLPCT_VALIDN:
+ return CTDT_LAYERCOL;
+
+ case CTSF_LAYERPCT_COUNT:
+ case CTSF_LAYERPCT_COUNT_RESPONSES:
+ case CTSF_LAYERPCT_RESPONSES:
+ case CTSF_LAYERPCT_RESPONSES_COUNT:
+ case CTSF_LAYERPCT_SUM:
+ case CTSF_LAYERPCT_TOTALN:
+ case CTSF_LAYERPCT_VALIDN:
+ return CTDT_LAYER;
+
+ case CTSF_LAYERROWPCT_COUNT:
+ case CTSF_LAYERROWPCT_COUNT_RESPONSES:
+ case CTSF_LAYERROWPCT_RESPONSES:
+ case CTSF_LAYERROWPCT_RESPONSES_COUNT:
+ case CTSF_LAYERROWPCT_SUM:
+ case CTSF_LAYERROWPCT_TOTALN:
+ case CTSF_LAYERROWPCT_VALIDN:
+ return CTDT_LAYERROW;
+
+ case CTSF_ROWPCT_COUNT:
+ case CTSF_ROWPCT_COUNT_RESPONSES:
+ case CTSF_ROWPCT_RESPONSES:
+ case CTSF_ROWPCT_RESPONSES_COUNT:
+ case CTSF_ROWPCT_SUM:
+ case CTSF_ROWPCT_TOTALN:
+ case CTSF_ROWPCT_VALIDN:
+ return CTDT_ROW;
+
+ case CTSF_SUBTABLEPCT_COUNT:
+ case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
+ case CTSF_SUBTABLEPCT_RESPONSES:
+ case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
+ case CTSF_SUBTABLEPCT_SUM:
+ case CTSF_SUBTABLEPCT_TOTALN:
+ case CTSF_SUBTABLEPCT_VALIDN:
+ return CTDT_SUBTABLE;
+
+ case CTSF_TABLEPCT_COUNT:
+ case CTSF_TABLEPCT_COUNT_RESPONSES:
+ case CTSF_TABLEPCT_RESPONSES:
+ case CTSF_TABLEPCT_RESPONSES_COUNT:
+ case CTSF_TABLEPCT_SUM:
+ case CTSF_TABLEPCT_TOTALN:
+ case CTSF_TABLEPCT_VALIDN:
+ return CTDT_TABLE;
+ }
+
+ NOT_REACHED ();
+}
+
static double
ctables_summary_value (const struct ctables_cell *cell,
union ctables_summary *s,
case CTSF_ECOUNT:
return s->valid;
- case CTSF_SUBTABLEPCT_COUNT:
- return cell->domains[CTDT_SUBTABLE]->valid ? s->valid / cell->domains[CTDT_SUBTABLE]->valid * 100 : SYSMIS;
-
case CTSF_ROWPCT_COUNT:
- return cell->domains[CTDT_ROW]->valid ? s->valid / cell->domains[CTDT_ROW]->valid * 100 : SYSMIS;
-
case CTSF_COLPCT_COUNT:
- return cell->domains[CTDT_COL]->valid ? s->valid / cell->domains[CTDT_COL]->valid * 100 : SYSMIS;
-
case CTSF_TABLEPCT_COUNT:
- return cell->domains[CTDT_TABLE]->valid ? s->valid / cell->domains[CTDT_TABLE]->valid * 100 : SYSMIS;
-
+ case CTSF_SUBTABLEPCT_COUNT:
case CTSF_LAYERPCT_COUNT:
- return cell->domains[CTDT_LAYER]->valid ? s->valid / cell->domains[CTDT_LAYER]->valid * 100 : SYSMIS;
-
case CTSF_LAYERROWPCT_COUNT:
- return cell->domains[CTDT_LAYERROW]->valid ? s->valid / cell->domains[CTDT_LAYERROW]->valid * 100 : SYSMIS;
-
case CTSF_LAYERCOLPCT_COUNT:
- return cell->domains[CTDT_LAYERCOL]->valid ? s->valid / cell->domains[CTDT_LAYERCOL]->valid * 100 : SYSMIS;
+ {
+ enum ctables_domain_type d = ctables_function_domain (ss->function);
+ return (cell->domains[d]->e_valid
+ ? s->valid / cell->domains[d]->e_valid * 100
+ : SYSMIS);
+ }
case CTSF_ROWPCT_VALIDN:
case CTSF_COLPCT_VALIDN:
ctables_cell_add__ (struct ctables_table *t, const struct ccase *c,
size_t ix[PIVOT_N_AXES],
const struct ctables_category *cats[PIVOT_N_AXES][10],
- double weight)
+ double d_weight, double e_weight)
{
struct ctables_cell *cell = ctables_cell_insert__ (t, c, ix, cats);
const struct ctables_nest *ss = &t->stacks[t->summary_axis].nests[ix[t->summary_axis]];
const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
for (size_t i = 0; i < specs->n; i++)
ctables_summary_add (&cell->summaries[i], &specs->specs[i], specs->var,
- case_data (c, specs->var), weight);
+ case_data (c, specs->var), d_weight, e_weight);
if (cell->contributes_to_domains)
- for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
- cell->domains[dt]->valid += weight;
+ {
+ for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
+ {
+ cell->domains[dt]->d_valid += d_weight;
+ cell->domains[dt]->e_valid += e_weight;
+ }
+ }
}
static void
recurse_totals (struct ctables_table *t, const struct ccase *c,
size_t ix[PIVOT_N_AXES],
const struct ctables_category *cats[PIVOT_N_AXES][10],
- double weight,
+ double d_weight, double e_weight,
enum pivot_axis_type start_axis, size_t start_nest)
{
for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
{
const struct ctables_category *save = cats[a][i];
cats[a][i] = total;
- ctables_cell_add__ (t, c, ix, cats, weight);
- recurse_totals (t, c, ix, cats, weight, a, i + 1);
+ ctables_cell_add__ (t, c, ix, cats, d_weight, e_weight);
+ recurse_totals (t, c, ix, cats, d_weight, e_weight, a, i + 1);
cats[a][i] = save;
}
}
recurse_subtotals (struct ctables_table *t, const struct ccase *c,
size_t ix[PIVOT_N_AXES],
const struct ctables_category *cats[PIVOT_N_AXES][10],
- double weight,
+ double d_weight, double e_weight,
enum pivot_axis_type start_axis, size_t start_nest)
{
for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
if (save->subtotal)
{
cats[a][i] = save->subtotal;
- ctables_cell_add__ (t, c, ix, cats, weight);
- recurse_subtotals (t, c, ix, cats, weight, a, i + 1);
+ ctables_cell_add__ (t, c, ix, cats, d_weight, e_weight);
+ recurse_subtotals (t, c, ix, cats, d_weight, e_weight, a, i + 1);
cats[a][i] = save;
}
}
ctables_cell_insert (struct ctables_table *t,
const struct ccase *c,
size_t ir, size_t ic, size_t il,
- double weight)
+ double d_weight, double e_weight)
{
size_t ix[PIVOT_N_AXES] = {
[PIVOT_AXIS_ROW] = ir,
}
}
- ctables_cell_add__ (t, c, ix, cats, weight);
+ ctables_cell_add__ (t, c, ix, cats, d_weight, e_weight);
- recurse_totals (t, c, ix, cats, weight, 0, 0);
- recurse_subtotals (t, c, ix, cats, weight, 0, 0);
+ recurse_totals (t, c, ix, cats, d_weight, e_weight, 0, 0);
+ recurse_subtotals (t, c, ix, cats, d_weight, e_weight, 0, 0);
}
struct merge_item
static bool
ctables_execute (struct dataset *ds, struct ctables *ct)
{
- struct casereader *input = casereader_create_filter_weight (proc_open (ds),
- dataset_dict (ds),
- NULL, NULL);
+ struct casereader *input = proc_open (ds);
bool warn_on_invalid = true;
- double total_weight = 0;
for (struct ccase *c = casereader_read (input); c;
case_unref (c), c = casereader_read (input))
{
- double weight = dict_get_case_weight (dataset_dict (ds), c,
- &warn_on_invalid);
- total_weight += weight;
+ double d_weight = dict_get_case_weight (dataset_dict (ds), c,
+ &warn_on_invalid);
+ double e_weight = (ct->e_weight
+ ? var_force_valid_weight (ct->e_weight,
+ case_num (c, ct->e_weight),
+ &warn_on_invalid)
+ : d_weight);
for (size_t i = 0; i < ct->n_tables; i++)
{
for (size_t ir = 0; ir < t->stacks[PIVOT_AXIS_ROW].n; ir++)
for (size_t ic = 0; ic < t->stacks[PIVOT_AXIS_COLUMN].n; ic++)
for (size_t il = 0; il < t->stacks[PIVOT_AXIS_LAYER].n; il++)
- ctables_cell_insert (t, c, ir, ic, il, weight);
+ ctables_cell_insert (t, c, ir, ic, il, d_weight, e_weight);
for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
if (t->label_axis[a] != a)
if (!lex_force_match_id (lexer, "VARIABLE"))
goto error;
lex_match (lexer, T_EQUALS);
- ct->base_weight = parse_variable (lexer, dataset_dict (ds));
- if (!ct->base_weight)
+ ct->e_weight = parse_variable (lexer, dataset_dict (ds));
+ if (!ct->e_weight)
goto error;
}
else if (lex_match_id (lexer, "HIDESMALLCOUNTS"))
projects / pspp / commitdiff