/* PSPP - linear regression. Copyright (C) 2005 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 2 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, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ #include #include #include #include #include #include #include "regression-export.h" #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include "gettext.h" #define REG_LARGE_DATA 1000 /* (headers) */ /* (specification) "REGRESSION" (regression_): *variables=custom; +statistics[st_]=r, coeff, anova, outs, zpp, label, sha, ci, bcov, ses, xtx, collin, tol, selection, f, defaults, all; export=custom; ^dependent=varlist; +save[sv_]=resid,pred; +method=enter. */ /* (declarations) */ /* (functions) */ static struct cmd_regression cmd; /* Moments for each of the variables used. */ struct moments_var { struct moments1 *m; struct variable *v; }; /* Linear regression models. */ static pspp_linreg_cache **models = NULL; /* Transformations for saving predicted values and residuals, etc. */ struct reg_trns { int n_trns; /* Number of transformations. */ int trns_id; /* Which trns is this one? */ pspp_linreg_cache *c; /* Linear model for this trns. */ }; /* Variables used (both explanatory and response). */ static struct variable **v_variables; /* Number of variables. */ static size_t n_variables; /* File where the model will be saved if the EXPORT subcommand is given. */ static struct file_handle *model_file; /* Return value for the procedure. */ static int pspp_reg_rc = CMD_SUCCESS; static bool run_regression (const struct ccase *, const struct casefile *, void *, const struct dataset *); /* STATISTICS subcommand output functions. */ static void reg_stats_r (pspp_linreg_cache *); static void reg_stats_coeff (pspp_linreg_cache *); static void reg_stats_anova (pspp_linreg_cache *); static void reg_stats_outs (pspp_linreg_cache *); static void reg_stats_zpp (pspp_linreg_cache *); static void reg_stats_label (pspp_linreg_cache *); static void reg_stats_sha (pspp_linreg_cache *); static void reg_stats_ci (pspp_linreg_cache *); static void reg_stats_f (pspp_linreg_cache *); static void reg_stats_bcov (pspp_linreg_cache *); static void reg_stats_ses (pspp_linreg_cache *); static void reg_stats_xtx (pspp_linreg_cache *); static void reg_stats_collin (pspp_linreg_cache *); static void reg_stats_tol (pspp_linreg_cache *); static void reg_stats_selection (pspp_linreg_cache *); static void statistics_keyword_output (void (*)(pspp_linreg_cache *), int, pspp_linreg_cache *); static void reg_stats_r (pspp_linreg_cache * c) { struct tab_table *t; int n_rows = 2; int n_cols = 5; double rsq; double adjrsq; double std_error; assert (c != NULL); rsq = c->ssm / c->sst; adjrsq = 1.0 - (1.0 - rsq) * (c->n_obs - 1.0) / (c->n_obs - c->n_indeps); std_error = sqrt ((c->n_indeps - 1.0) / (c->n_obs - 1.0)); t = tab_create (n_cols, n_rows, 0); tab_dim (t, tab_natural_dimensions); tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1); tab_hline (t, TAL_2, 0, n_cols - 1, 1); tab_vline (t, TAL_2, 2, 0, n_rows - 1); tab_vline (t, TAL_0, 1, 0, 0); tab_text (t, 1, 0, TAB_CENTER | TAT_TITLE, _("R")); tab_text (t, 2, 0, TAB_CENTER | TAT_TITLE, _("R Square")); tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("Adjusted R Square")); tab_text (t, 4, 0, TAB_CENTER | TAT_TITLE, _("Std. Error of the Estimate")); tab_float (t, 1, 1, TAB_RIGHT, sqrt (rsq), 10, 2); tab_float (t, 2, 1, TAB_RIGHT, rsq, 10, 2); tab_float (t, 3, 1, TAB_RIGHT, adjrsq, 10, 2); tab_float (t, 4, 1, TAB_RIGHT, std_error, 10, 2); tab_title (t, _("Model Summary")); tab_submit (t); } /* Table showing estimated regression coefficients. */ static void reg_stats_coeff (pspp_linreg_cache * c) { size_t j; int n_cols = 7; int n_rows; double t_stat; double pval; double coeff; double std_err; double beta; const char *label; char *tmp; const struct variable *v; const union value *val; const char *val_s; struct tab_table *t; assert (c != NULL); tmp = xnmalloc (MAX_STRING, sizeof (*tmp)); n_rows = c->n_coeffs + 2; t = tab_create (n_cols, n_rows, 0); tab_headers (t, 2, 0, 1, 0); tab_dim (t, tab_natural_dimensions); tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1); tab_hline (t, TAL_2, 0, n_cols - 1, 1); tab_vline (t, TAL_2, 2, 0, n_rows - 1); tab_vline (t, TAL_0, 1, 0, 0); tab_text (t, 2, 0, TAB_CENTER | TAT_TITLE, _("B")); tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("Std. Error")); tab_text (t, 4, 0, TAB_CENTER | TAT_TITLE, _("Beta")); tab_text (t, 5, 0, TAB_CENTER | TAT_TITLE, _("t")); tab_text (t, 6, 0, TAB_CENTER | TAT_TITLE, _("Significance")); tab_text (t, 1, 1, TAB_LEFT | TAT_TITLE, _("(Constant)")); coeff = c->coeff[0]->estimate; tab_float (t, 2, 1, 0, coeff, 10, 2); std_err = sqrt (gsl_matrix_get (c->cov, 0, 0)); tab_float (t, 3, 1, 0, std_err, 10, 2); beta = coeff / c->depvar_std; tab_float (t, 4, 1, 0, beta, 10, 2); t_stat = coeff / std_err; tab_float (t, 5, 1, 0, t_stat, 10, 2); pval = 2 * gsl_cdf_tdist_Q (fabs (t_stat), 1.0); tab_float (t, 6, 1, 0, pval, 10, 2); for (j = 1; j <= c->n_indeps; j++) { v = pspp_coeff_get_var (c->coeff[j], 0); label = var_to_string (v); /* Do not overwrite the variable's name. */ strncpy (tmp, label, MAX_STRING); if (var_is_alpha (v)) { /* Append the value associated with this coefficient. This makes sense only if we us the usual binary encoding for that value. */ val = pspp_coeff_get_value (c->coeff[j], v); val_s = var_get_value_name (v, val); strncat (tmp, val_s, MAX_STRING); } tab_text (t, 1, j + 1, TAB_CENTER, tmp); /* Regression coefficients. */ coeff = c->coeff[j]->estimate; tab_float (t, 2, j + 1, 0, coeff, 10, 2); /* Standard error of the coefficients. */ std_err = sqrt (gsl_matrix_get (c->cov, j, j)); tab_float (t, 3, j + 1, 0, std_err, 10, 2); /* 'Standardized' coefficient, i.e., regression coefficient if all variables had unit variance. */ beta = gsl_vector_get (c->indep_std, j); beta *= coeff / c->depvar_std; tab_float (t, 4, j + 1, 0, beta, 10, 2); /* Test statistic for H0: coefficient is 0. */ t_stat = coeff / std_err; tab_float (t, 5, j + 1, 0, t_stat, 10, 2); /* P values for the test statistic above. */ pval = 2 * gsl_cdf_tdist_Q (fabs (t_stat), (double) (c->n_obs - c->n_coeffs)); tab_float (t, 6, j + 1, 0, pval, 10, 2); } tab_title (t, _("Coefficients")); tab_submit (t); free (tmp); } /* Display the ANOVA table. */ static void reg_stats_anova (pspp_linreg_cache * c) { int n_cols = 7; int n_rows = 4; const double msm = c->ssm / c->dfm; const double mse = c->sse / c->dfe; const double F = msm / mse; const double pval = gsl_cdf_fdist_Q (F, c->dfm, c->dfe); struct tab_table *t; assert (c != NULL); t = tab_create (n_cols, n_rows, 0); tab_headers (t, 2, 0, 1, 0); tab_dim (t, tab_natural_dimensions); tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1); tab_hline (t, TAL_2, 0, n_cols - 1, 1); tab_vline (t, TAL_2, 2, 0, n_rows - 1); tab_vline (t, TAL_0, 1, 0, 0); tab_text (t, 2, 0, TAB_CENTER | TAT_TITLE, _("Sum of Squares")); tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("df")); tab_text (t, 4, 0, TAB_CENTER | TAT_TITLE, _("Mean Square")); tab_text (t, 5, 0, TAB_CENTER | TAT_TITLE, _("F")); tab_text (t, 6, 0, TAB_CENTER | TAT_TITLE, _("Significance")); tab_text (t, 1, 1, TAB_LEFT | TAT_TITLE, _("Regression")); tab_text (t, 1, 2, TAB_LEFT | TAT_TITLE, _("Residual")); tab_text (t, 1, 3, TAB_LEFT | TAT_TITLE, _("Total")); /* Sums of Squares */ tab_float (t, 2, 1, 0, c->ssm, 10, 2); tab_float (t, 2, 3, 0, c->sst, 10, 2); tab_float (t, 2, 2, 0, c->sse, 10, 2); /* Degrees of freedom */ tab_float (t, 3, 1, 0, c->dfm, 4, 0); tab_float (t, 3, 2, 0, c->dfe, 4, 0); tab_float (t, 3, 3, 0, c->dft, 4, 0); /* Mean Squares */ tab_float (t, 4, 1, TAB_RIGHT, msm, 8, 3); tab_float (t, 4, 2, TAB_RIGHT, mse, 8, 3); tab_float (t, 5, 1, 0, F, 8, 3); tab_float (t, 6, 1, 0, pval, 8, 3); tab_title (t, _("ANOVA")); tab_submit (t); } static void reg_stats_outs (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_zpp (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_label (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_sha (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_ci (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_f (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_bcov (pspp_linreg_cache * c) { int n_cols; int n_rows; int i; int k; int row; int col; const char *label; struct tab_table *t; assert (c != NULL); n_cols = c->n_indeps + 1 + 2; n_rows = 2 * (c->n_indeps + 1); t = tab_create (n_cols, n_rows, 0); tab_headers (t, 2, 0, 1, 0); tab_dim (t, tab_natural_dimensions); tab_box (t, TAL_2, TAL_2, -1, TAL_1, 0, 0, n_cols - 1, n_rows - 1); tab_hline (t, TAL_2, 0, n_cols - 1, 1); tab_vline (t, TAL_2, 2, 0, n_rows - 1); tab_vline (t, TAL_0, 1, 0, 0); tab_text (t, 0, 0, TAB_CENTER | TAT_TITLE, _("Model")); tab_text (t, 1, 1, TAB_CENTER | TAT_TITLE, _("Covariances")); for (i = 1; i < c->n_coeffs; i++) { const struct variable *v = pspp_coeff_get_var (c->coeff[i], 0); label = var_to_string (v); tab_text (t, 2, i, TAB_CENTER, label); tab_text (t, i + 2, 0, TAB_CENTER, label); for (k = 1; k < c->n_coeffs; k++) { col = (i <= k) ? k : i; row = (i <= k) ? i : k; tab_float (t, k + 2, i, TAB_CENTER, gsl_matrix_get (c->cov, row, col), 8, 3); } } tab_title (t, _("Coefficient Correlations")); tab_submit (t); } static void reg_stats_ses (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_xtx (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_collin (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_tol (pspp_linreg_cache * c) { assert (c != NULL); } static void reg_stats_selection (pspp_linreg_cache * c) { assert (c != NULL); } static void statistics_keyword_output (void (*function) (pspp_linreg_cache *), int keyword, pspp_linreg_cache * c) { if (keyword) { (*function) (c); } } static void subcommand_statistics (int *keywords, pspp_linreg_cache * c) { /* The order here must match the order in which the STATISTICS keywords appear in the specification section above. */ enum { r, coeff, anova, outs, zpp, label, sha, ci, bcov, ses, xtx, collin, tol, selection, f, defaults, all }; int i; int d = 1; if (keywords[all]) { /* Set everything but F. */ for (i = 0; i < f; i++) { keywords[i] = 1; } } else { for (i = 0; i < all; i++) { if (keywords[i]) { d = 0; } } /* Default output: ANOVA table, parameter estimates, and statistics for variables not entered into model, if appropriate. */ if (keywords[defaults] | d) { keywords[anova] = 1; keywords[outs] = 1; keywords[coeff] = 1; keywords[r] = 1; } } statistics_keyword_output (reg_stats_r, keywords[r], c); statistics_keyword_output (reg_stats_anova, keywords[anova], c); statistics_keyword_output (reg_stats_coeff, keywords[coeff], c); statistics_keyword_output (reg_stats_outs, keywords[outs], c); statistics_keyword_output (reg_stats_zpp, keywords[zpp], c); statistics_keyword_output (reg_stats_label, keywords[label], c); statistics_keyword_output (reg_stats_sha, keywords[sha], c); statistics_keyword_output (reg_stats_ci, keywords[ci], c); statistics_keyword_output (reg_stats_f, keywords[f], c); statistics_keyword_output (reg_stats_bcov, keywords[bcov], c); statistics_keyword_output (reg_stats_ses, keywords[ses], c); statistics_keyword_output (reg_stats_xtx, keywords[xtx], c); statistics_keyword_output (reg_stats_collin, keywords[collin], c); statistics_keyword_output (reg_stats_tol, keywords[tol], c); statistics_keyword_output (reg_stats_selection, keywords[selection], c); } /* Free the transformation. Free its linear model if this transformation is the last one. */ static bool regression_trns_free (void *t_) { bool result = true; struct reg_trns *t = t_; if (t->trns_id == t->n_trns) { result = pspp_linreg_cache_free (t->c); } free (t); return result; } /* Gets the predicted values. */ static int regression_trns_pred_proc (void *t_, struct ccase *c, casenumber case_idx UNUSED) { size_t i; size_t n_vals; struct reg_trns *trns = t_; pspp_linreg_cache *model; union value *output = NULL; const union value **vals = NULL; struct variable **vars = NULL; assert (trns != NULL); model = trns->c; assert (model != NULL); assert (model->depvar != NULL); assert (model->pred != NULL); vars = xnmalloc (model->n_coeffs, sizeof (*vars)); n_vals = (*model->get_vars) (model, vars); vals = xnmalloc (n_vals, sizeof (*vals)); output = case_data_rw (c, model->pred); assert (output != NULL); for (i = 0; i < n_vals; i++) { vals[i] = case_data (c, vars[i]); } output->f = (*model->predict) ((const struct variable **) vars, vals, model, n_vals); free (vals); free (vars); return TRNS_CONTINUE; } /* Gets the residuals. */ static int regression_trns_resid_proc (void *t_, struct ccase *c, casenumber case_idx UNUSED) { size_t i; size_t n_vals; struct reg_trns *trns = t_; pspp_linreg_cache *model; union value *output = NULL; const union value **vals = NULL; const union value *obs = NULL; struct variable **vars = NULL; assert (trns != NULL); model = trns->c; assert (model != NULL); assert (model->depvar != NULL); assert (model->resid != NULL); vars = xnmalloc (model->n_coeffs, sizeof (*vars)); n_vals = (*model->get_vars) (model, vars); vals = xnmalloc (n_vals, sizeof (*vals)); output = case_data_rw (c, model->resid); assert (output != NULL); for (i = 0; i < n_vals; i++) { vals[i] = case_data (c, vars[i]); } obs = case_data (c, model->depvar); output->f = (*model->residual) ((const struct variable **) vars, vals, obs, model, n_vals); free (vals); free (vars); return TRNS_CONTINUE; } /* Returns false if NAME is a duplicate of any existing variable name. */ static bool try_name (const struct dictionary *dict, const char *name) { if (dict_lookup_var (dict, name) != NULL) return false; return true; } static void reg_get_name (const struct dictionary *dict, char name[LONG_NAME_LEN], const char prefix[LONG_NAME_LEN]) { int i = 1; snprintf (name, LONG_NAME_LEN, "%s%d", prefix, i); while (!try_name (dict, name)) { i++; snprintf (name, LONG_NAME_LEN, "%s%d", prefix, i); } } static void reg_save_var (struct dataset *ds, const char *prefix, trns_proc_func * f, pspp_linreg_cache * c, struct variable **v, int n_trns) { struct dictionary *dict = dataset_dict (ds); static int trns_index = 1; char name[LONG_NAME_LEN]; struct variable *new_var; struct reg_trns *t = NULL; t = xmalloc (sizeof (*t)); t->trns_id = trns_index; t->n_trns = n_trns; t->c = c; reg_get_name (dict, name, prefix); new_var = dict_create_var (dict, name, 0); assert (new_var != NULL); *v = new_var; add_transformation (ds, f, regression_trns_free, t); trns_index++; } static void subcommand_save (struct dataset *ds, int save, pspp_linreg_cache ** models) { pspp_linreg_cache **lc; int n_trns = 0; int i; assert (models != NULL); if (save) { /* Count the number of transformations we will need. */ for (i = 0; i < REGRESSION_SV_count; i++) { if (cmd.a_save[i]) { n_trns++; } } n_trns *= cmd.n_dependent; for (lc = models; lc < models + cmd.n_dependent; lc++) { assert (*lc != NULL); assert ((*lc)->depvar != NULL); if (cmd.a_save[REGRESSION_SV_RESID]) { reg_save_var (ds, "RES", regression_trns_resid_proc, *lc, &(*lc)->resid, n_trns); } if (cmd.a_save[REGRESSION_SV_PRED]) { reg_save_var (ds, "PRED", regression_trns_pred_proc, *lc, &(*lc)->pred, n_trns); } } } else { for (lc = models; lc < models + cmd.n_dependent; lc++) { assert (*lc != NULL); pspp_linreg_cache_free (*lc); } } } static int reg_inserted (const struct variable *v, struct variable **varlist, int n_vars) { int i; for (i = 0; i < n_vars; i++) { if (v == varlist[i]) { return 1; } } return 0; } static void reg_print_categorical_encoding (FILE * fp, pspp_linreg_cache * c) { int i; int n_vars = 0; struct variable **varlist; fprintf (fp, "%s", reg_export_categorical_encode_1); varlist = xnmalloc (c->n_indeps, sizeof (*varlist)); for (i = 1; i < c->n_indeps; i++) /* c->coeff[0] is the intercept. */ { struct pspp_coeff *coeff = c->coeff[i]; const struct variable *v = pspp_coeff_get_var (coeff, 0); if (var_is_alpha (v)) { if (!reg_inserted (v, varlist, n_vars)) { fprintf (fp, "struct pspp_reg_categorical_variable %s;\n\t", var_get_name (v)); varlist[n_vars] = (struct variable *) v; n_vars++; } } } fprintf (fp, "int n_vars = %d;\n\t", n_vars); fprintf (fp, "struct pspp_reg_categorical_variable *varlist[%d] = {", n_vars); for (i = 0; i < n_vars - 1; i++) { fprintf (fp, "&%s,\n\t\t", var_get_name (varlist[i])); } fprintf (fp, "&%s};\n\t", var_get_name (varlist[i])); for (i = 0; i < n_vars; i++) { int n_categories = cat_get_n_categories (varlist[i]); int j; fprintf (fp, "%s.name = \"%s\";\n\t", var_get_name (varlist[i]), var_get_name (varlist[i])); fprintf (fp, "%s.n_vals = %d;\n\t", var_get_name (varlist[i]), n_categories); for (j = 0; j < n_categories; j++) { union value *val = cat_subscript_to_value (j, varlist[i]); fprintf (fp, "%s.values[%d] = \"%s\";\n\t", var_get_name (varlist[i]), j, var_get_value_name (varlist[i], val)); } } fprintf (fp, "%s", reg_export_categorical_encode_2); } static void reg_print_depvars (FILE * fp, pspp_linreg_cache * c) { int i; struct pspp_coeff *coeff; const struct variable *v; fprintf (fp, "char *model_depvars[%d] = {", c->n_indeps); for (i = 1; i < c->n_indeps; i++) { coeff = c->coeff[i]; v = pspp_coeff_get_var (coeff, 0); fprintf (fp, "\"%s\",\n\t\t", var_get_name (v)); } coeff = c->coeff[i]; v = pspp_coeff_get_var (coeff, 0); fprintf (fp, "\"%s\"};\n\t", var_get_name (v)); } static void reg_print_getvar (FILE * fp, pspp_linreg_cache * c) { fprintf (fp, "static int\npspp_reg_getvar (char *v_name)\n{\n\t"); fprintf (fp, "int i;\n\tint n_vars = %d;\n\t", c->n_indeps); reg_print_depvars (fp, c); fprintf (fp, "for (i = 0; i < n_vars; i++)\n\t{\n\t\t"); fprintf (fp, "if (strncmp (v_name, model_depvars[i], PSPP_REG_MAXLEN) == 0)\n\t\t{\n\t\t\t"); fprintf (fp, "return i;\n\t\t}\n\t}\n}\n"); } static int reg_has_categorical (pspp_linreg_cache * c) { int i; const struct variable *v; for (i = 1; i < c->n_coeffs; i++) { v = pspp_coeff_get_var (c->coeff[i], 0); if (var_is_alpha (v)) return 1; } return 0; } static void subcommand_export (int export, pspp_linreg_cache * c) { FILE *fp; size_t i; size_t j; int n_quantiles = 100; double tmp; struct pspp_coeff *coeff; if (export) { assert (c != NULL); assert (model_file != NULL); fp = fopen (fh_get_file_name (model_file), "w"); assert (fp != NULL); fprintf (fp, "%s", reg_preamble); reg_print_getvar (fp, c); if (reg_has_categorical (c)) { reg_print_categorical_encoding (fp, c); } fprintf (fp, "%s", reg_export_t_quantiles_1); for (i = 0; i < n_quantiles - 1; i++) { tmp = 0.5 + 0.005 * (double) i; fprintf (fp, "%.15e,\n\t\t", gsl_cdf_tdist_Pinv (tmp, c->n_obs - c->n_indeps)); } fprintf (fp, "%.15e};\n\t", gsl_cdf_tdist_Pinv (.9995, c->n_obs - c->n_indeps)); fprintf (fp, "%s", reg_export_t_quantiles_2); fprintf (fp, "%s", reg_mean_cmt); fprintf (fp, "double\npspp_reg_estimate (const double *var_vals,"); fprintf (fp, "const char *var_names[])\n{\n\t"); fprintf (fp, "double model_coeffs[%d] = {", c->n_indeps); for (i = 1; i < c->n_indeps; i++) { coeff = c->coeff[i]; fprintf (fp, "%.15e,\n\t\t", coeff->estimate); } coeff = c->coeff[i]; fprintf (fp, "%.15e};\n\t", coeff->estimate); coeff = c->coeff[0]; fprintf (fp, "double estimate = %.15e;\n\t", coeff->estimate); fprintf (fp, "int i;\n\tint j;\n\n\t"); fprintf (fp, "for (i = 0; i < %d; i++)\n\t", c->n_indeps); fprintf (fp, "%s", reg_getvar); fprintf (fp, "const double cov[%d][%d] = {\n\t", c->n_coeffs, c->n_coeffs); for (i = 0; i < c->cov->size1 - 1; i++) { fprintf (fp, "{"); for (j = 0; j < c->cov->size2 - 1; j++) { fprintf (fp, "%.15e, ", gsl_matrix_get (c->cov, i, j)); } fprintf (fp, "%.15e},\n\t", gsl_matrix_get (c->cov, i, j)); } fprintf (fp, "{"); for (j = 0; j < c->cov->size2 - 1; j++) { fprintf (fp, "%.15e, ", gsl_matrix_get (c->cov, c->cov->size1 - 1, j)); } fprintf (fp, "%.15e}\n\t", gsl_matrix_get (c->cov, c->cov->size1 - 1, c->cov->size2 - 1)); fprintf (fp, "};\n\tint n_vars = %d;\n\tint i;\n\tint j;\n\t", c->n_indeps); fprintf (fp, "double unshuffled_vals[%d];\n\t", c->n_indeps); fprintf (fp, "%s", reg_variance); fprintf (fp, "%s", reg_export_confidence_interval); tmp = c->mse * c->mse; fprintf (fp, "%s %.15e", reg_export_prediction_interval_1, tmp); fprintf (fp, "%s %.15e", reg_export_prediction_interval_2, tmp); fprintf (fp, "%s", reg_export_prediction_interval_3); fclose (fp); fp = fopen ("pspp_model_reg.h", "w"); fprintf (fp, "%s", reg_header); fclose (fp); } } static int regression_custom_export (struct lexer *lexer, struct dataset *ds UNUSED, struct cmd_regression *cmd UNUSED, void *aux UNUSED) { /* 0 on failure, 1 on success, 2 on failure that should result in syntax error */ if (!lex_force_match (lexer, '(')) return 0; if (lex_match (lexer, '*')) model_file = NULL; else { model_file = fh_parse (lexer, FH_REF_FILE); if (model_file == NULL) return 0; } if (!lex_force_match (lexer, ')')) return 0; return 1; } int cmd_regression (struct lexer *lexer, struct dataset *ds) { if (!parse_regression (lexer, ds, &cmd, NULL)) return CMD_FAILURE; models = xnmalloc (cmd.n_dependent, sizeof *models); if (!multipass_procedure_with_splits (ds, run_regression, &cmd)) return CMD_CASCADING_FAILURE; subcommand_save (ds, cmd.sbc_save, models); free (v_variables); free (models); return pspp_reg_rc; } /* Is variable k the dependent variable? */ static bool is_depvar (size_t k, const struct variable *v) { return v == v_variables[k]; } /* Mark missing cases. Return the number of non-missing cases. Compute the first two moments. */ static size_t mark_missing_cases (const struct casefile *cf, struct variable *v, int *is_missing_case, double n_data, struct moments_var *mom) { struct casereader *r; struct ccase c; size_t row; const union value *val; double w = 1.0; for (r = casefile_get_reader (cf, NULL); casereader_read (r, &c); case_destroy (&c)) { row = casereader_cnum (r) - 1; val = case_data (&c, v); if (mom != NULL) { moments1_add (mom->m, val->f, w); } cat_value_update (v, val); if (var_is_value_missing (v, val, MV_ANY)) { if (!is_missing_case[row]) { /* Now it is missing. */ n_data--; is_missing_case[row] = 1; } } } casereader_destroy (r); return n_data; } /* Parser for the variables sub command */ static int regression_custom_variables (struct lexer *lexer, struct dataset *ds, struct cmd_regression *cmd UNUSED, void *aux UNUSED) { const struct dictionary *dict = dataset_dict (ds); lex_match (lexer, '='); if ((lex_token (lexer) != T_ID || dict_lookup_var (dict, lex_tokid (lexer)) == NULL) && lex_token (lexer) != T_ALL) return 2; if (!parse_variables (lexer, dict, &v_variables, &n_variables, PV_NONE)) { free (v_variables); return 0; } assert (n_variables); return 1; } /* Count the explanatory variables. The user may or may not have specified a response variable in the syntax. */ static int get_n_indep (const struct variable *v) { int result; int i = 0; result = n_variables; while (i < n_variables) { if (is_depvar (i, v)) { result--; i = n_variables; } i++; } return result; } /* Read from the active file. Identify the explanatory variables in v_variables. Encode categorical variables. Drop cases with missing values. */ static int prepare_data (int n_data, int is_missing_case[], struct variable **indep_vars, struct variable *depvar, const struct casefile *cf, struct moments_var *mom) { int i; int j; assert (indep_vars != NULL); j = 0; for (i = 0; i < n_variables; i++) { if (!is_depvar (i, depvar)) { indep_vars[j] = v_variables[i]; j++; if (var_is_alpha (v_variables[i])) { /* Make a place to hold the binary vectors corresponding to this variable's values. */ cat_stored_values_create (v_variables[i]); } n_data = mark_missing_cases (cf, v_variables[i], is_missing_case, n_data, mom + i); } } /* Mark missing cases for the dependent variable. */ n_data = mark_missing_cases (cf, depvar, is_missing_case, n_data, NULL); return n_data; } static void coeff_init (pspp_linreg_cache * c, struct design_matrix *dm) { c->coeff = xnmalloc (dm->m->size2 + 1, sizeof (*c->coeff)); c->coeff[0] = xmalloc (sizeof (*(c->coeff[0]))); /* The first coefficient is the intercept. */ c->coeff[0]->v_info = NULL; /* Intercept has no associated variable. */ pspp_coeff_init (c->coeff + 1, dm); } /* Put the moments in the linreg cache. */ static void compute_moments (pspp_linreg_cache *c, struct moments_var *mom, struct design_matrix *dm, size_t n) { size_t i; size_t j; double weight; double mean; double variance; double skewness; double kurtosis; /* Scan the variable names in the columns of the design matrix. When we find the variable we need, insert its mean in the cache. */ for (i = 0; i < dm->m->size2; i++) { for (j = 0; j < n; j++) { if (design_matrix_col_to_var (dm, i) == (mom + j)->v) { moments1_calculate ((mom + j)->m, &weight, &mean, &variance, &skewness, &kurtosis); gsl_vector_set (c->indep_means, i, mean); gsl_vector_set (c->indep_std, i, sqrt (variance)); } } } } static bool run_regression (const struct ccase *first, const struct casefile *cf, void *cmd_ UNUSED, const struct dataset *ds) { size_t i; size_t n_data = 0; /* Number of valide cases. */ size_t n_cases; /* Number of cases. */ size_t row; size_t case_num; int n_indep = 0; int k; /* Keep track of the missing cases. */ int *is_missing_case; const union value *val; struct casereader *r; struct ccase c; struct variable **indep_vars; struct design_matrix *X; struct moments_var *mom; gsl_vector *Y; pspp_linreg_opts lopts; assert (models != NULL); output_split_file_values (ds, first); if (!v_variables) { dict_get_vars (dataset_dict (ds), &v_variables, &n_variables, 1u << DC_SYSTEM); } n_cases = casefile_get_case_cnt (cf); for (i = 0; i < cmd.n_dependent; i++) { if (!var_is_numeric (cmd.v_dependent[i])) { msg (SE, gettext ("Dependent variable must be numeric.")); pspp_reg_rc = CMD_FAILURE; return true; } } is_missing_case = xnmalloc (n_cases, sizeof (*is_missing_case)); mom = xnmalloc (n_variables, sizeof (*mom)); for (i = 0; i < n_variables; i++) { (mom + i)->m = moments1_create (MOMENT_VARIANCE); (mom + i)->v = v_variables[i]; } lopts.get_depvar_mean_std = 1; for (k = 0; k < cmd.n_dependent; k++) { n_indep = get_n_indep ((const struct variable *) cmd.v_dependent[k]); lopts.get_indep_mean_std = xnmalloc (n_indep, sizeof (int)); indep_vars = xnmalloc (n_indep, sizeof *indep_vars); assert (indep_vars != NULL); for (i = 0; i < n_cases; i++) { is_missing_case[i] = 0; } n_data = prepare_data (n_cases, is_missing_case, indep_vars, cmd.v_dependent[k], (const struct casefile *) cf, mom); Y = gsl_vector_alloc (n_data); X = design_matrix_create (n_indep, (const struct variable **) indep_vars, n_data); for (i = 0; i < X->m->size2; i++) { lopts.get_indep_mean_std[i] = 1; } models[k] = pspp_linreg_cache_alloc (X->m->size1, X->m->size2); models[k]->indep_means = gsl_vector_alloc (X->m->size2); models[k]->indep_std = gsl_vector_alloc (X->m->size2); models[k]->depvar = (const struct variable *) cmd.v_dependent[k]; /* For large data sets, use QR decomposition. */ if (n_data > sqrt (n_indep) && n_data > REG_LARGE_DATA) { models[k]->method = PSPP_LINREG_SVD; } /* The second pass fills the design matrix. */ row = 0; for (r = casefile_get_reader (cf, NULL); casereader_read (r, &c); case_destroy (&c)) /* Iterate over the cases. */ { case_num = casereader_cnum (r) - 1; if (!is_missing_case[case_num]) { for (i = 0; i < n_variables; ++i) /* Iterate over the variables for the current case. */ { val = case_data (&c, v_variables[i]); /* Independent/dependent variable separation. The 'variables' subcommand specifies a varlist which contains both dependent and independent variables. The dependent variables are specified with the 'dependent' subcommand, and maybe also in the 'variables' subcommand. We need to separate the two. */ if (!is_depvar (i, cmd.v_dependent[k])) { if (var_is_alpha (v_variables[i])) { design_matrix_set_categorical (X, row, v_variables[i], val); } else { design_matrix_set_numeric (X, row, v_variables[i], val); } } } val = case_data (&c, cmd.v_dependent[k]); gsl_vector_set (Y, row, val->f); row++; } } /* Now that we know the number of coefficients, allocate space and store pointers to the variables that correspond to the coefficients. */ coeff_init (models[k], X); /* Find the least-squares estimates and other statistics. */ pspp_linreg ((const gsl_vector *) Y, X->m, &lopts, models[k]); compute_moments (models[k], mom, X, n_variables); subcommand_statistics (cmd.a_statistics, models[k]); subcommand_export (cmd.sbc_export, models[k]); gsl_vector_free (Y); design_matrix_destroy (X); free (indep_vars); free (lopts.get_indep_mean_std); casereader_destroy (r); } for (i = 0; i < n_variables; i++) { moments1_destroy ((mom + i)->m); } free (mom); free (is_missing_case); return true; } /* Local Variables: mode: c End: */