static gboolean
analysis_tool_principal_components_engine_run (data_analysis_output_t *dao,
analysis_tools_data_generic_t *info)
{
int l = g_slist_length (info->input), i;
GSList *inputdata;
GnmFunc *fd_mean;
GnmFunc *fd_var;
GnmFunc *fd_eigen;
GnmFunc *fd_mmult;
GnmFunc *fd_munit;
GnmFunc *fd_sqrt;
GnmFunc *fd_count;
GnmFunc *fd_sum;
GnmFunc *fd_and;
GnmFunc *fd_if;
GnmExpr const *expr;
GnmExpr const *expr_count;
GnmExpr const *expr_munit;
GnmExpr const *expr_and;
int data_points;
GnmExprList *and_args = NULL;
if (!dao_cell_is_visible (dao, l, 9 + 3 * l)) {
dao_set_bold (dao, 0, 0, 0, 0);
dao_set_italic (dao, 0, 0, 0, 0);
dao_set_cell (dao, 0, 0,
_("Principal components analysis has "
"insufficient space."));
return 0;
}
fd_mean = gnm_func_lookup_or_add_placeholder
("AVERAGE", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_mean);
fd_var = gnm_func_lookup_or_add_placeholder
("VAR", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_var);
fd_eigen = gnm_func_lookup_or_add_placeholder
("EIGEN", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_eigen);
fd_mmult = gnm_func_lookup_or_add_placeholder
("MMULT", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_mmult);
fd_munit = gnm_func_lookup_or_add_placeholder
("MUNIT", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_munit);
fd_sqrt = gnm_func_lookup_or_add_placeholder
("SQRT", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_sqrt);
fd_count = gnm_func_lookup_or_add_placeholder
("COUNT", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_count);
fd_sum = gnm_func_lookup_or_add_placeholder
("SUM", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_sum);
fd_and = gnm_func_lookup_or_add_placeholder
("AND", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_and);
fd_if = gnm_func_lookup_or_add_placeholder
("IF", dao->sheet ? dao->sheet->workbook : NULL, FALSE);
gnm_func_ref (fd_if);
dao_set_bold (dao, 0, 0, 0, 0);
dao_set_italic (dao, 0, 0, 0, 11 + 3 * l);
dao_set_format (dao, 0, 0, 0, 0,
_("\"Principal Components Analysis\";"
"[Red]\"Principal Components Analysis is invalid.\""));
dao_set_align (dao, 0, 0, 0, 0,
HALIGN_LEFT, VALIGN_BOTTOM);
dao->offset_row++;
analysis_tool_table (dao, info, _("Covariances:"), "COVAR", TRUE);
dao->offset_row--;
for (i = 1, inputdata = info->input; inputdata != NULL; i++, inputdata = inputdata->next)
analysis_tools_write_label (inputdata->data, dao, info, 0, 9 + 2 * l + i, i);
data_points = value_area_get_width (info->input->data, NULL) *
value_area_get_height (info->input->data, NULL);
for (i = 0; i < l; i++)
and_args = gnm_expr_list_prepend
(and_args,
gnm_expr_new_binary
(gnm_expr_new_constant (value_new_int (data_points)),
GNM_EXPR_OP_EQUAL,
make_cellref (1 + i, 3 + l)));
expr_and = gnm_expr_new_funcall (fd_and, and_args);
dao_set_cell_expr (dao, 0, 0,
gnm_expr_new_funcall3
(fd_if,
expr_and,
gnm_expr_new_constant (value_new_int (1)),
gnm_expr_new_constant (value_new_int (-1))));
dao_set_merge (dao,0,0,2,0);
set_cell_text_col (dao, 0, 3 + l,
_("/Count:"
"/Mean:"
"/Variance:"
"//Eigenvalues:"
"/Eigenvectors:"));
dao_set_cell (dao, 0, 11 + 3 * l, _("Percent of Trace:"));
dao_set_italic (dao, 0, 9 + 2 * l, 1 + l, 9 + 2 * l);
dao_set_percent (dao, 1, 11 + 3 * l, 1 + l, 11 + 3 * l);
for (i = 1, inputdata = info->input; inputdata != NULL; i++, inputdata = inputdata->next) {
expr = gnm_expr_new_constant (value_dup (inputdata->data));
dao_set_cell_expr (dao, i, 3 + l,
gnm_expr_new_funcall1 (fd_count, gnm_expr_copy (expr)));
dao_set_cell_expr (dao, i, 4 + l,
gnm_expr_new_funcall1 (fd_mean, gnm_expr_copy (expr)));
dao_set_cell_expr (dao, i, 5 + l,
gnm_expr_new_funcall1 (fd_var, expr));
}
expr_count = gnm_expr_new_binary (make_cellref (0,-4), GNM_EXPR_OP_DIV,
gnm_expr_new_binary (make_cellref (0,-4), GNM_EXPR_OP_SUB,
gnm_expr_new_constant (value_new_int (1))));
expr = gnm_expr_new_funcall1
(fd_eigen, gnm_expr_new_binary
(expr_count, GNM_EXPR_OP_MULT, make_rangeref (0, - (5 + l), l - 1, - 6)));
dao_set_array_expr (dao, 1, 7 + l, l, l + 1, expr);
for (i = 1; i <= l; i++) {
dao_set_align (dao, i, 9 + 2 * l, i, 9 + 2 * l,
HALIGN_CENTER, VALIGN_BOTTOM);
dao_set_cell_printf (dao, i, 9 + 2 * l, "\xce\xbe%i", i);
dao_set_cell_expr (dao, i, 11 + 3 * l,
gnm_expr_new_binary (make_cellref (0,- 4 - 2 * l),
GNM_EXPR_OP_DIV,
gnm_expr_new_funcall1
(fd_sum,
dao_get_rangeref (dao, 1, 7 + l, l, 7 + l))));
}
expr_munit = gnm_expr_new_funcall1 (fd_munit, gnm_expr_new_constant (value_new_int (l)));
expr = gnm_expr_new_funcall2 (fd_mmult,
gnm_expr_new_binary
(gnm_expr_new_funcall1
(fd_sqrt, gnm_expr_new_binary
(gnm_expr_new_constant (value_new_int (1)),
GNM_EXPR_OP_DIV,
make_rangeref (0, - 5 - l, l - 1, - 5 - l))),
GNM_EXPR_OP_MULT,
gnm_expr_copy (expr_munit)),
make_rangeref (0, - 2 - l, l - 1, - 3));
expr = gnm_expr_new_funcall2 (fd_mmult, expr,
gnm_expr_new_binary
(gnm_expr_new_funcall1
(fd_sqrt, make_rangeref (0, - 3 - l, l - 1, - 3 - l)),
GNM_EXPR_OP_MULT,
expr_munit));
dao_set_array_expr (dao, 1, 10 + 2 * l, l, l, expr);
gnm_func_unref (fd_mean);
gnm_func_unref (fd_var);
gnm_func_unref (fd_eigen);
gnm_func_unref (fd_mmult);
gnm_func_unref (fd_munit);
gnm_func_unref (fd_sqrt);
gnm_func_unref (fd_count);
gnm_func_unref (fd_sum);
gnm_func_unref (fd_and);
gnm_func_unref (fd_if);
dao_redraw_respan (dao);
return 0;
}
static gboolean
analysis_tool_wilcoxon_mann_whitney_engine_run (data_analysis_output_t *dao,
analysis_tools_data_generic_b_t *info)
{
GnmFunc *fd_count;
GnmFunc *fd_sum;
GnmFunc *fd_rows;
GnmFunc *fd_rank_avg;
GnmFunc *fd_rank;
GnmFunc *fd_min;
GnmFunc *fd_normdist;
GnmFunc *fd_sqrt;
GnmFunc *fd_if;
GnmFunc *fd_isblank;
GnmExpr const *expr_total;
GnmExpr const *expr_pop_1;
GnmExpr const *expr_pop_2;
GnmExpr const *expr_u;
GnmExpr const *expr_count_total;
GnmValue *val_1 = value_dup (info->range_1);
GnmValue *val_2 = value_dup (info->range_2);
Workbook *wb = dao->sheet ? dao->sheet->workbook : NULL;
fd_count = gnm_func_lookup_or_add_placeholder ("COUNT");
gnm_func_ref (fd_count);
fd_sum = gnm_func_lookup_or_add_placeholder ("SUM");
gnm_func_ref (fd_sum);
fd_rows = gnm_func_lookup_or_add_placeholder ("ROWS");
gnm_func_ref (fd_rows);
fd_rank_avg = gnm_func_lookup_or_add_placeholder ("RANK.AVG");
gnm_func_ref (fd_rank_avg);
fd_rank = gnm_func_lookup_or_add_placeholder ("RANK");
gnm_func_ref (fd_rank);
fd_min = gnm_func_lookup_or_add_placeholder ("MIN");
gnm_func_ref (fd_min);
fd_normdist = gnm_func_lookup_or_add_placeholder ("NORMDIST");
gnm_func_ref (fd_normdist);
fd_sqrt = gnm_func_lookup_or_add_placeholder ("SQRT");
gnm_func_ref (fd_sqrt);
fd_if = gnm_func_lookup_or_add_placeholder ("IF");
gnm_func_ref (fd_if);
fd_isblank = gnm_func_lookup_or_add_placeholder ("ISBLANK");
gnm_func_ref (fd_isblank);
dao_set_italic (dao, 0, 0, 0, 8);
dao_set_italic (dao, 0, 1, 3, 1);
dao_set_merge (dao, 0, 0, 3, 0);
dao_set_cell (dao, 0, 0, _("Wilcoxon-Mann-Whitney Test"));
set_cell_text_col (dao, 0, 2, _("/Rank-Sum"
"/N"
"/U"
"/Ties"
"/Statistic"
"/U-Statistic"
"/p-Value"));
dao_set_cell (dao, 3, 1, _("Total"));
/* Label */
analysis_tools_write_label_ftest (val_1, dao, 1, 1, info->labels, 1);
analysis_tools_write_label_ftest (val_2, dao, 2, 1, info->labels, 2);
expr_total = analysis_tool_combine_area (val_1, val_2, wb);
expr_pop_1 = gnm_expr_new_constant (val_1);
expr_pop_2 = gnm_expr_new_constant (val_2);
/* =sum(if(isblank(region1),0,rank.avg(region1,combined_regions,1))) */
dao_set_cell_array_expr (dao, 1, 2,
gnm_expr_new_funcall1
(fd_sum,
gnm_expr_new_funcall3
(fd_if,
gnm_expr_new_funcall1
(fd_isblank,
gnm_expr_copy (expr_pop_1)),
gnm_expr_new_constant (value_new_int (0)),
gnm_expr_new_funcall3
(fd_rank_avg,
gnm_expr_copy (expr_pop_1),
gnm_expr_copy (expr_total),
gnm_expr_new_constant (value_new_int (1))))));
dao_set_cell_array_expr (dao, 2, 2,
gnm_expr_new_funcall1
(fd_sum,
gnm_expr_new_funcall3
(fd_if,
gnm_expr_new_funcall1
(fd_isblank,
gnm_expr_copy (expr_pop_2)),
gnm_expr_new_constant (value_new_int (0)),
gnm_expr_new_funcall3
(fd_rank_avg,
gnm_expr_copy (expr_pop_2),
gnm_expr_copy (expr_total),
gnm_expr_new_constant (value_new_int (1))))));
expr_count_total = gnm_expr_new_funcall1
(fd_count, gnm_expr_copy (expr_total));
dao_set_cell_expr (dao, 3, 2,
gnm_expr_new_binary
(gnm_expr_new_binary
(gnm_expr_copy (expr_count_total),
GNM_EXPR_OP_MULT,
gnm_expr_new_binary
(gnm_expr_copy (expr_count_total),
GNM_EXPR_OP_ADD,
gnm_expr_new_constant (value_new_int (1)))),
GNM_EXPR_OP_DIV,
gnm_expr_new_constant (value_new_int (2))));
dao_set_cell_expr (dao, 1, 3,
gnm_expr_new_funcall1
(fd_count,
expr_pop_1));
dao_set_cell_expr (dao, 2, 3,
gnm_expr_new_funcall1
(fd_count,
expr_pop_2));
dao_set_cell_expr (dao, 3, 3,
gnm_expr_new_funcall1
(fd_count,
gnm_expr_copy (expr_total)));
expr_u = gnm_expr_new_binary
(make_cellref (0,- 2), GNM_EXPR_OP_SUB,
gnm_expr_new_binary
(gnm_expr_new_binary
(make_cellref (0,- 1),
GNM_EXPR_OP_MULT,
gnm_expr_new_binary
(make_cellref (0,- 1),
GNM_EXPR_OP_ADD,
gnm_expr_new_constant (value_new_int (1)))),
GNM_EXPR_OP_DIV,
gnm_expr_new_constant (value_new_int (2))));
dao_set_cell_expr (dao, 1, 4, gnm_expr_copy (expr_u));
dao_set_cell_expr (dao, 2, 4, expr_u);
dao_set_cell_expr (dao, 3, 4,
gnm_expr_new_binary
(make_cellref (-2,-1),
GNM_EXPR_OP_MULT,
make_cellref (-1,-1)));
dao_set_cell_array_expr (dao, 1, 5,
gnm_expr_new_funcall1
(fd_sum,
gnm_expr_new_binary
(gnm_expr_new_funcall2
(fd_rank_avg,
gnm_expr_copy (expr_total),
gnm_expr_copy (expr_total)),
GNM_EXPR_OP_SUB,
gnm_expr_new_funcall2
(fd_rank,
gnm_expr_copy (expr_total),
gnm_expr_copy (expr_total)))));
if (dao_cell_is_visible (dao, 2, 4)) {
GnmExpr const *expr_prod;
GnmExpr const *expr_sqrt;
GnmExpr const *expr_normdist;
expr_prod = gnm_expr_new_binary
(make_cellref (0,-5),
GNM_EXPR_OP_MULT,
make_cellref (1,-5));
expr_sqrt = gnm_expr_new_funcall1
(fd_sqrt,
gnm_expr_new_binary
(gnm_expr_new_binary
(gnm_expr_copy(expr_prod),
GNM_EXPR_OP_MULT,
gnm_expr_new_binary
(gnm_expr_new_binary
(make_cellref (0,-5),
GNM_EXPR_OP_ADD,
make_cellref (1,-5)),
GNM_EXPR_OP_ADD,
gnm_expr_new_constant (value_new_int (1)))),
GNM_EXPR_OP_DIV,
gnm_expr_new_constant (value_new_int (12))));
expr_normdist = gnm_expr_new_funcall4
(fd_normdist,
make_cellref (0,-1),
gnm_expr_new_binary
(expr_prod,
GNM_EXPR_OP_DIV,
gnm_expr_new_constant (value_new_int (2))),
expr_sqrt,
gnm_expr_new_constant (value_new_bool (TRUE)));
dao_set_cell_expr (dao, 1, 6,
gnm_expr_new_funcall2
(fd_min,
make_cellref (0,-4),
make_cellref (1,-4)));
dao_set_cell_expr (dao, 1, 7,
gnm_expr_new_funcall2
(fd_min,
make_cellref (0,-3),
make_cellref (1,-3)));
dao_set_cell_expr (dao, 1, 8,
gnm_expr_new_binary
(gnm_expr_new_constant (value_new_int (2)),
GNM_EXPR_OP_MULT,
expr_normdist));
dao_set_cell_comment (dao, 1, 8,
_("This p-value is calculated using a\n"
"normal approximation, so it is\n"
"only valid for large samples of\n"
"at least 15 observations in each\n"
"population, and few if any ties."));
} else {
dao_set_cell_na (dao, 1, 6);
dao_set_cell_comment (dao, 1, 6,
_("Since there is insufficient space\n"
"for the third column of output,\n"
"this value is not calculated."));
dao_set_cell_na (dao, 1, 7);
dao_set_cell_comment (dao, 1, 7,
_("Since there is insufficient space\n"
"for the third column of output,\n"
"this value is not calculated."));
dao_set_cell_na (dao, 1, 8);
dao_set_cell_comment (dao, 1, 8,
_("Since there is insufficient space\n"
"for the third column of output,\n"
"this value is not calculated."));
}
gnm_expr_free (expr_count_total);
gnm_expr_free (expr_total);
gnm_func_unref (fd_count);
gnm_func_unref (fd_sum);
gnm_func_unref (fd_rows);
gnm_func_unref (fd_rank_avg);
gnm_func_unref (fd_rank);
gnm_func_unref (fd_min);
gnm_func_unref (fd_normdist);
gnm_func_unref (fd_sqrt);
gnm_func_unref (fd_if);
gnm_func_unref (fd_isblank);
dao_redraw_respan (dao);
return 0;
}
