From 09907ebe926441d6a6834f358844e10358ea86e2 Mon Sep 17 00:00:00 2001 From: Jason Stover Date: Tue, 11 Mar 2008 03:28:48 +0000 Subject: [PATCH] Removed use of coefficient 0 as intercept; removed subcommand EXPORT. --- src/language/stats/ChangeLog | 5 + src/language/stats/regression.q | 276 +++----------------------------- src/math/ChangeLog | 5 + src/math/coefficient.c | 14 +- src/math/linreg/ChangeLog | 4 + src/math/linreg/linreg.c | 15 +- src/math/linreg/linreg.h | 4 +- src/math/linreg/predict.c | 5 +- 8 files changed, 50 insertions(+), 278 deletions(-) diff --git a/src/language/stats/ChangeLog b/src/language/stats/ChangeLog index ae4110d1..0f32d774 100644 --- a/src/language/stats/ChangeLog +++ b/src/language/stats/ChangeLog @@ -1,3 +1,8 @@ +2008-03-10 Jason Stover + + * regression.q (run_regression): Removed code for EXPORT + subcommand. Remove use of coefficient 0 as the intercept. + 2008-02-14 John Darrington * examine.q: Fixed counts of missing variables. Thanks to diff --git a/src/language/stats/regression.q b/src/language/stats/regression.q index bed4f79d..e7ced45c 100644 --- a/src/language/stats/regression.q +++ b/src/language/stats/regression.q @@ -22,7 +22,6 @@ #include #include -#include "regression-export.h" #include #include #include @@ -75,7 +74,6 @@ f, defaults, all; - export=custom; ^dependent=varlist; +save[sv_]=resid,pred; +method=enter. @@ -113,12 +111,6 @@ static const struct variable **v_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; - static bool run_regression (struct casereader *, struct cmd_regression *, struct dataset *, pspp_linreg_cache **); @@ -185,6 +177,7 @@ reg_stats_coeff (pspp_linreg_cache * c) size_t j; int n_cols = 7; int n_rows; + int this_row; double t_stat; double pval; double coeff; @@ -197,7 +190,7 @@ reg_stats_coeff (pspp_linreg_cache * c) struct tab_table *t; assert (c != NULL); - n_rows = c->n_coeffs + 2; + n_rows = c->n_coeffs + 3; t = tab_create (n_cols, n_rows, 0); tab_headers (t, 2, 0, 1, 0); @@ -213,18 +206,18 @@ reg_stats_coeff (pspp_linreg_cache * c) 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); + tab_float (t, 2, 1, 0, c->intercept, 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; + beta = c->intercept / c->depvar_std; tab_float (t, 4, 1, 0, beta, 10, 2); - t_stat = coeff / std_err; + t_stat = c->intercept / 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++) + for (j = 0; j < c->n_coeffs; j++) { + this_row = j + 2; struct string tstr; ds_init_empty (&tstr); @@ -245,37 +238,37 @@ reg_stats_coeff (pspp_linreg_cache * c) var_append_value_name (v, val, &tstr); } - tab_text (t, 1, j + 1, TAB_CENTER, ds_cstr (&tstr)); + tab_text (t, 1, this_row, TAB_CENTER, ds_cstr (&tstr)); /* Regression coefficients. */ coeff = c->coeff[j]->estimate; - tab_float (t, 2, j + 1, 0, coeff, 10, 2); + tab_float (t, 2, this_row, 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); + std_err = sqrt (gsl_matrix_get (c->cov, j + 1, j + 1)); + tab_float (t, 3, this_row, 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 = gsl_vector_get (c->indep_std, j + 1); beta *= coeff / c->depvar_std; - tab_float (t, 4, j + 1, 0, beta, 10, 2); + tab_float (t, 4, this_row, 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); + tab_float (t, 5, this_row, 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_float (t, 6, this_row, 0, pval, 10, 2); ds_destroy (&tstr); } tab_title (t, _("Coefficients")); @@ -395,7 +388,7 @@ reg_stats_bcov (pspp_linreg_cache * c) 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++) + for (i = 0; i < c->n_coeffs; i++) { const struct variable *v = pspp_coeff_get_var (c->coeff[i], 0); label = var_to_string (v); @@ -672,7 +665,6 @@ reg_save_var (struct dataset *ds, const char *prefix, trns_proc_func * f, add_transformation (ds, f, regression_trns_free, t); trns_index++; } - static void subcommand_save (struct dataset *ds, int save, pspp_linreg_cache ** models) { @@ -722,232 +714,6 @@ subcommand_save (struct dataset *ds, int save, pspp_linreg_cache ** models) } } -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++) - { - struct string vstr; - const union value *val = cat_subscript_to_value (j, varlist[i]); - ds_init_empty (&vstr); - var_append_value_name (varlist[i], val, &vstr); - fprintf (fp, "%s.values[%d] = \"%s\";\n\t", - var_get_name (varlist[i]), j, - ds_cstr (&vstr)); - - ds_destroy (&vstr); - } - } - 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 - { - fh_unref (model_file); - 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) { @@ -957,10 +723,8 @@ cmd_regression (struct lexer *lexer, struct dataset *ds) bool ok; size_t i; - model_file = NULL; if (!parse_regression (lexer, ds, &cmd, NULL)) { - fh_unref (model_file); return CMD_FAILURE; } @@ -981,7 +745,6 @@ cmd_regression (struct lexer *lexer, struct dataset *ds) free (v_variables); free (models); free_regression (&cmd); - fh_unref (model_file); return ok ? CMD_SUCCESS : CMD_FAILURE; } @@ -1094,10 +857,8 @@ prepare_categories (struct casereader *input, 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); + c->coeff = xnmalloc (dm->m->size2, sizeof (*c->coeff)); + pspp_coeff_init (c->coeff, dm); } /* @@ -1255,7 +1016,6 @@ run_regression (struct casereader *input, struct cmd_regression *cmd, if (!taint_has_tainted_successor (casereader_get_taint (input))) { subcommand_statistics (cmd->a_statistics, models[k]); - subcommand_export (cmd->sbc_export, models[k]); } gsl_vector_free (Y); diff --git a/src/math/ChangeLog b/src/math/ChangeLog index 48b30627..1d3b95bb 100644 --- a/src/math/ChangeLog +++ b/src/math/ChangeLog @@ -1,3 +1,8 @@ +2008-03-10 Jason Stover + + * coefficient.c (pspp_linreg_get_coeff): Removed use of + coefficient 0 as intercept. + 2008-03-01 Jason Stover * coefficient.c (pspp_coeff_init): Ensure first arg is not a null diff --git a/src/math/coefficient.c b/src/math/coefficient.c index 1868bda4..156c782d 100644 --- a/src/math/coefficient.c +++ b/src/math/coefficient.c @@ -195,11 +195,8 @@ pspp_linreg_get_coeff (const pspp_linreg_cache * c, { return NULL; } - /* - C->N_COEFFS == 1 means regression through the origin. - */ - i = (c->n_coeffs > 1) ? 1 : 0; - result = c->coeff[i]; + i = 0; + result = c->coeff[0]; tmp = pspp_coeff_get_var (result, 0); while (tmp != v && i < c->n_coeffs) { @@ -207,8 +204,11 @@ pspp_linreg_get_coeff (const pspp_linreg_cache * c, tmp = pspp_coeff_get_var (result, 0); i++; } - if (i >= c->n_coeffs) + if (tmp != v) { + /* + Not found. + */ return NULL; } if (var_is_numeric (v)) @@ -229,7 +229,7 @@ pspp_linreg_get_coeff (const pspp_linreg_cache * c, result = c->coeff[i]; tmp = pspp_coeff_get_var (result, 0); } - if (i == c->n_coeffs) + if (i == c->n_coeffs && tmp != v) { return NULL; } diff --git a/src/math/linreg/ChangeLog b/src/math/linreg/ChangeLog index 552d093b..c92db77a 100644 --- a/src/math/linreg/ChangeLog +++ b/src/math/linreg/ChangeLog @@ -1,3 +1,7 @@ +2008-03-10 Jason Stover + + * linreg.c (pspp_linreg): Remove use of coefficient 0 as intercept. + 2008-03-08 Jason Stover * linreg.c (pspp_linreg_get_vars): Clean up the loop that searches diff --git a/src/math/linreg/linreg.c b/src/math/linreg/linreg.c index 9465875e..217e201c 100644 --- a/src/math/linreg/linreg.c +++ b/src/math/linreg/linreg.c @@ -107,10 +107,7 @@ pspp_linreg_get_vars (const void *c_, const struct variable **v) { v[i] = NULL; } - /* - Start at c->coeff[1] to avoid the intercept. - */ - for (j = 1; j < c->n_coeffs; j++) + for (j = 0; j < c->n_coeffs; j++) { tmp = pspp_coeff_get_var (c->coeff[j], 0); assert (tmp != NULL); @@ -243,9 +240,9 @@ pspp_linreg (const gsl_vector * Y, const gsl_matrix * X, cache->dft = cache->n_obs - 1; cache->dfm = cache->n_indeps; cache->dfe = cache->dft - cache->dfm; - cache->n_coeffs = X->size2 + 1; /* Adjust this later to allow for - regression through the origin. - */ + cache->n_coeffs = X->size2; + cache->intercept = 0.0; + if (cache->method == PSPP_LINREG_SWEEP) { gsl_matrix *sw; @@ -319,7 +316,7 @@ pspp_linreg (const gsl_vector * Y, const gsl_matrix * X, for (i = 0; i < cache->n_indeps; i++) { tmp = gsl_matrix_get (sw, i, cache->n_indeps); - cache->coeff[i + 1]->estimate = tmp; + cache->coeff[i]->estimate = tmp; m -= tmp * gsl_vector_get (cache->indep_means, i); } /* @@ -355,7 +352,7 @@ pspp_linreg (const gsl_vector * Y, const gsl_matrix * X, } gsl_matrix_set (cache->cov, 0, 0, tmp); - cache->coeff[0]->estimate = m; + cache->intercept = m; } else { diff --git a/src/math/linreg/linreg.h b/src/math/linreg/linreg.h index d09752a2..a7d408af 100644 --- a/src/math/linreg/linreg.h +++ b/src/math/linreg/linreg.h @@ -95,7 +95,8 @@ struct pspp_linreg_cache_struct { int n_obs; /* Number of observations. */ int n_indeps; /* Number of independent variables. */ - int n_coeffs; + int n_coeffs; /* The intercept is not considered a + coefficient here. */ /* The variable struct is ignored during estimation. It is here so @@ -105,6 +106,7 @@ struct pspp_linreg_cache_struct gsl_vector *residuals; struct pspp_coeff **coeff; + double intercept; int method; /* Method to use to estimate parameters. */ /* Means and standard deviations of the variables. diff --git a/src/math/linreg/predict.c b/src/math/linreg/predict.c index 8eba054a..91184687 100644 --- a/src/math/linreg/predict.c +++ b/src/math/linreg/predict.c @@ -51,7 +51,7 @@ pspp_linreg_predict (const struct variable **predictors, { const pspp_linreg_cache *c = c_; int j; - size_t next_coef = 1; + size_t next_coef = 0; const struct pspp_coeff **coef_list; const struct pspp_coeff *coe; double result; @@ -67,8 +67,7 @@ pspp_linreg_predict (const struct variable **predictors, return c->depvar_mean; } coef_list = xnmalloc (c->n_coeffs, sizeof (*coef_list)); - *coef_list = c->coeff[0]; - result = c->coeff[0]->estimate; /* Intercept. */ + result = c->intercept; /* The loops guard against the possibility that the caller passed us -- 2.30.2