From 63aa70751714736d6577c590484ad926492b2fe2 Mon Sep 17 00:00:00 2001 From: John Darrington Date: Thu, 4 May 2017 17:31:37 +0200 Subject: [PATCH] FACTOR: New subcommand: MATRIX IN --- NEWS | 2 + doc/statistics.texi | 22 +++- src/language/stats/factor.c | 214 +++++++++++++++++++++++++----------- 3 files changed, 169 insertions(+), 69 deletions(-) diff --git a/NEWS b/NEWS index dcdfb90bfe..0390e74f33 100644 --- a/NEWS +++ b/NEWS @@ -6,6 +6,8 @@ Please send PSPP bug reports to bug-gnu-pspp@gnu.org. Changes from 0.10.2 to 0.10.4: + * The FACTOR command can now analyse matrix files prepared with MATRIX DATA. + * The MATRIX DATA command has been added. * Some inappropriate properties in selection dialogs have been corrected. diff --git a/doc/statistics.texi b/doc/statistics.texi index e18c6109db..b5026e7b2c 100644 --- a/doc/statistics.texi +++ b/doc/statistics.texi @@ -785,7 +785,10 @@ Fixes for any of these deficiencies would be welcomed. @cindex data reduction @display -FACTOR VARIABLES=@var{var_list} +FACTOR @{ + VARIABLES=@var{var_list}, + MATRIX IN (@{CORR,COV@}=@{*,@var{file_spec}@}) + @} [ /METHOD = @{CORRELATION, COVARIANCE@} ] @@ -809,10 +812,21 @@ FACTOR VARIABLES=@var{var_list} The @cmd{FACTOR} command performs Factor Analysis or Principal Axis Factoring on a dataset. It may be used to find common factors in the data or for data reduction purposes. -The @subcmd{VARIABLES} subcommand is required. It lists the variables -which are to partake in the analysis. (The @subcmd{ANALYSIS} +The @subcmd{VARIABLES} subcommand is required (unless the @subcmd{MATRIX IN} +subcommand is used). +It lists the variables which are to partake in the analysis. (The @subcmd{ANALYSIS} subcommand may optionally further limit the variables that -participate; it is not useful and implemented only for compatibility.) +participate; it is useful primarily in conjunction with @subcmd{MATRIX IN}.) + +If @subcmd{MATRIX IN} instead of @subcmd{VARIABLES} is specified, then the analysis +is performed on a pre-prepared correlation or covariance matrix file instead of on +individual data cases. Typically the matrix file will have been generated by +@cmd{MATRIX DATA} (@pxref{MATRIX DATA}) or provided by a third party. +If specified, @subcmd{MATRIX IN} must be followed by @samp{COV} or @samp{CORR}, +then by @samp{=} and @var{file_spec} all in parentheses. +@var{file_spec} may either be an asterisk, which indicates the currently loaded +dataset, or it may be a filename to be loaded. @xref{MATRIX DATA} for the expected +format of the file. The @subcmd{/EXTRACTION} subcommand is used to specify the way in which factors (components) are extracted from the data. If @subcmd{PC} is specified, then Principal Components Analysis is used. diff --git a/src/language/stats/factor.c b/src/language/stats/factor.c index 472c340fd7..3043fced29 100644 --- a/src/language/stats/factor.c +++ b/src/language/stats/factor.c @@ -1,5 +1,6 @@ /* PSPP - a program for statistical analysis. - Copyright (C) 2009, 2010, 2011, 2012, 2014, 2015, 2016 Free Software Foundation, Inc. + Copyright (C) 2009, 2010, 2011, 2012, 2014, 2015, + 2016, 2017 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 @@ -24,6 +25,7 @@ #include #include +#include "data/any-reader.h" #include "data/casegrouper.h" #include "data/casereader.h" #include "data/casewriter.h" @@ -35,6 +37,8 @@ #include "language/lexer/lexer.h" #include "language/lexer/value-parser.h" #include "language/lexer/variable-parser.h" +#include "language/data-io/file-handle.h" +#include "language/data-io/matrix-reader.h" #include "libpspp/cast.h" #include "libpspp/message.h" #include "libpspp/misc.h" @@ -45,6 +49,7 @@ #include "output/charts/scree.h" #include "output/tab.h" + #include "gettext.h" #define _(msgid) gettext (msgid) #define N_(msgid) msgid @@ -221,13 +226,13 @@ struct cmd_factor bool sort; }; + struct idata { /* Intermediate values used in calculation */ + struct matrix_material mm; - const gsl_matrix *corr ; /* The correlation matrix */ - gsl_matrix *cov ; /* The covariance matrix */ - const gsl_matrix *n ; /* Matrix of number of samples */ + gsl_matrix *analysis_matrix; /* A pointer to either mm.corr or mm.cov */ gsl_vector *eval ; /* The eigenvalues */ gsl_matrix *evec ; /* The eigenvectors */ @@ -237,6 +242,8 @@ struct idata gsl_vector *msr ; /* Multiple Squared Regressions */ double detR; /* The determinant of the correlation matrix */ + + struct covariance *cvm; }; static struct idata * @@ -259,10 +266,10 @@ idata_free (struct idata *id) gsl_vector_free (id->msr); gsl_vector_free (id->eval); gsl_matrix_free (id->evec); - if (id->cov != NULL) - gsl_matrix_free (id->cov); - if (id->corr != NULL) - gsl_matrix_free (CONST_CAST (gsl_matrix *, id->corr)); + if (id->mm.cov != NULL) + gsl_matrix_free (id->mm.cov); + if (id->mm.corr != NULL) + gsl_matrix_free (CONST_CAST (gsl_matrix *, id->mm.corr)); free (id); } @@ -1002,11 +1009,14 @@ iterate_factor_matrix (const gsl_matrix *r, gsl_vector *communalities, gsl_matri static bool run_factor (struct dataset *ds, const struct cmd_factor *factor); +static void do_factor_by_matrix (const struct cmd_factor *factor, struct idata *idata); + + int cmd_factor (struct lexer *lexer, struct dataset *ds) { - const struct dictionary *dict = dataset_dict (ds); + struct dictionary *dict = NULL; int n_iterations = 25; struct cmd_factor factor; factor.n_vars = 0; @@ -1026,26 +1036,76 @@ cmd_factor (struct lexer *lexer, struct dataset *ds) factor.sort = false; factor.plot = 0; factor.rotation = ROT_VARIMAX; + factor.wv = NULL; factor.rconverge = 0.0001; - factor.wv = dict_get_weight (dict); - lex_match (lexer, T_SLASH); - if (!lex_force_match_id (lexer, "VARIABLES")) + struct matrix_reader *mr = NULL; + struct casereader *matrix_reader = NULL; + + if (lex_match_id (lexer, "VARIABLES")) { - goto error; + lex_match (lexer, T_EQUALS); + dict = dataset_dict (ds); + factor.wv = dict_get_weight (dict); + + if (!parse_variables_const (lexer, dict, &factor.vars, &factor.n_vars, + PV_NO_DUPLICATE | PV_NUMERIC)) + goto error; } + else if (lex_match_id (lexer, "MATRIX")) + { + if (! lex_force_match_id (lexer, "IN")) + goto error; + if (!lex_force_match (lexer, T_LPAREN)) + { + goto error; + } + if (lex_match_id (lexer, "CORR")) + { + } + else if (lex_match_id (lexer, "COV")) + { + } + else + { + lex_error (lexer, _("Matrix input for %s must be either COV or CORR"), "FACTOR"); + goto error; + } + if (! lex_force_match (lexer, T_EQUALS)) + goto error; + if (lex_match (lexer, T_ASTERISK)) + { + dict = dataset_dict (ds); + matrix_reader = casereader_clone (dataset_source (ds)); + } + else + { + struct file_handle *fh = fh_parse (lexer, FH_REF_FILE, NULL); + if (fh == NULL) + goto error; + + matrix_reader + = any_reader_open_and_decode (fh, NULL, &dict, NULL); - lex_match (lexer, T_EQUALS); + if (! (matrix_reader && dict)) + { + goto error; + } + } - if (!parse_variables_const (lexer, dict, &factor.vars, &factor.n_vars, - PV_NO_DUPLICATE | PV_NUMERIC)) - goto error; + if (! lex_force_match (lexer, T_RPAREN)) + goto error; - if (factor.n_vars < 2) - msg (MW, _("Factor analysis on a single variable is not useful.")); + mr = create_matrix_reader_from_case_reader (dict, matrix_reader, + &factor.vars, &factor.n_vars); + } + else + { + goto error; + } while (lex_token (lexer) != T_ENDCMD) { @@ -1408,13 +1468,35 @@ cmd_factor (struct lexer *lexer, struct dataset *ds) if ( factor.rotation == ROT_NONE ) factor.print &= ~PRINT_ROTATION; - if ( ! run_factor (ds, &factor)) - goto error; + if (factor.n_vars < 2) + msg (MW, _("Factor analysis on a single variable is not useful.")); + + if (matrix_reader) + { + struct idata *id = idata_alloc (factor.n_vars); + + while (next_matrix_from_reader (&id->mm, mr, + factor.vars, factor.n_vars)) + { + do_factor_by_matrix (&factor, id); + + id->mm.corr = NULL; + id->mm.cov = NULL; + } + + idata_free (id); + } + else + if ( ! run_factor (ds, &factor)) + goto error; + + destroy_matrix_reader (mr); free (factor.vars); return CMD_SUCCESS; error: + destroy_matrix_reader (mr); free (factor.vars); return CMD_FAILURE; } @@ -1983,7 +2065,7 @@ show_correlation_matrix (const struct cmd_factor *factor, const struct idata *id for (i = 0; i < factor->n_vars; ++i) { for (j = 0; j < factor->n_vars; ++j) - tab_double (t, heading_columns + i, y + j, 0, gsl_matrix_get (idata->corr, i, j), NULL, RC_OTHER); + tab_double (t, heading_columns + i, y + j, 0, gsl_matrix_get (idata->mm.corr, i, j), NULL, RC_OTHER); } } @@ -1996,8 +2078,8 @@ show_correlation_matrix (const struct cmd_factor *factor, const struct idata *id { for (j = 0; j < factor->n_vars; ++j) { - double rho = gsl_matrix_get (idata->corr, i, j); - double w = gsl_matrix_get (idata->n, i, j); + double rho = gsl_matrix_get (idata->mm.corr, i, j); + double w = gsl_matrix_get (idata->mm.n, i, j); if (i == j) continue; @@ -2019,58 +2101,61 @@ show_correlation_matrix (const struct cmd_factor *factor, const struct idata *id } - static void do_factor (const struct cmd_factor *factor, struct casereader *r) { struct ccase *c; - const gsl_matrix *var_matrix; - const gsl_matrix *mean_matrix; - - const gsl_matrix *analysis_matrix; struct idata *idata = idata_alloc (factor->n_vars); - struct covariance *cov = covariance_1pass_create (factor->n_vars, factor->vars, + idata->cvm = covariance_1pass_create (factor->n_vars, factor->vars, factor->wv, factor->exclude); for ( ; (c = casereader_read (r) ); case_unref (c)) { - covariance_accumulate (cov, c); + covariance_accumulate (idata->cvm, c); } - idata->cov = covariance_calculate (cov); + idata->mm.cov = covariance_calculate (idata->cvm); - if (idata->cov == NULL) + if (idata->mm.cov == NULL) { msg (MW, _("The dataset contains no complete observations. No analysis will be performed.")); - covariance_destroy (cov); + covariance_destroy (idata->cvm); goto finish; } - var_matrix = covariance_moments (cov, MOMENT_VARIANCE); - mean_matrix = covariance_moments (cov, MOMENT_MEAN); - idata->n = covariance_moments (cov, MOMENT_NONE); + idata->mm.var_matrix = covariance_moments (idata->cvm, MOMENT_VARIANCE); + idata->mm.mean_matrix = covariance_moments (idata->cvm, MOMENT_MEAN); + idata->mm.n = covariance_moments (idata->cvm, MOMENT_NONE); + do_factor_by_matrix (factor, idata); - if ( factor->method == METHOD_CORR) - { - idata->corr = correlation_from_covariance (idata->cov, var_matrix); + finish: + idata_free (idata); + casereader_destroy (r); +} - analysis_matrix = idata->corr; - } +static void +do_factor_by_matrix (const struct cmd_factor *factor, struct idata *idata) +{ + if (idata->mm.cov && !idata->mm.corr) + idata->mm.corr = correlation_from_covariance (idata->mm.cov, idata->mm.var_matrix); + if (idata->mm.corr && !idata->mm.cov) + idata->mm.cov = covariance_from_correlation (idata->mm.corr, idata->mm.var_matrix); + if (factor->method == METHOD_CORR) + idata->analysis_matrix = idata->mm.corr; else - analysis_matrix = idata->cov; - + idata->analysis_matrix = idata->mm.cov; if (factor->print & PRINT_DETERMINANT || factor->print & PRINT_KMO) { int sign = 0; - const int size = idata->corr->size1; + const int size = idata->mm.corr->size1; gsl_permutation *p = gsl_permutation_calloc (size); gsl_matrix *tmp = gsl_matrix_calloc (size, size); - gsl_matrix_memcpy (tmp, idata->corr); + gsl_matrix_memcpy (tmp, idata->mm.corr); gsl_linalg_LU_decomp (tmp, p, &sign); idata->detR = gsl_linalg_LU_det (tmp, sign); @@ -2121,9 +2206,9 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) const struct variable *v = factor->vars[i]; tab_text (t, 0, i + heading_rows, TAB_LEFT | TAT_TITLE, var_to_string (v)); - tab_double (t, 1, i + heading_rows, 0, gsl_matrix_get (mean_matrix, i, i), NULL, RC_OTHER); - tab_double (t, 2, i + heading_rows, 0, sqrt (gsl_matrix_get (var_matrix, i, i)), NULL, RC_OTHER); - tab_double (t, 3, i + heading_rows, 0, gsl_matrix_get (idata->n, i, i), NULL, RC_WEIGHT); + tab_double (t, 1, i + heading_rows, 0, gsl_matrix_get (idata->mm.mean_matrix, i, i), NULL, RC_OTHER); + tab_double (t, 2, i + heading_rows, 0, sqrt (gsl_matrix_get (idata->mm.var_matrix, i, i)), NULL, RC_OTHER); + tab_double (t, 3, i + heading_rows, 0, gsl_matrix_get (idata->mm.n, i, i), NULL, RC_WEIGHT); } tab_submit (t); @@ -2135,7 +2220,7 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) double sum_ssq_r = 0; double sum_ssq_a = 0; - double df = factor->n_vars * ( factor->n_vars - 1) / 2; + double df = factor->n_vars * (factor->n_vars - 1) / 2; double w = 0; @@ -2153,7 +2238,7 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) struct tab_table *t = tab_create (nc, nr); tab_title (t, _("KMO and Bartlett's Test")); - x = clone_matrix (idata->corr); + x = clone_matrix (idata->mm.corr); gsl_linalg_cholesky_decomp (x); gsl_linalg_cholesky_invert (x); @@ -2194,9 +2279,9 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) missing values are involved. The best thing I can think of is to take the mean average. */ w = 0; - for (i = 0; i < idata->n->size1; ++i) - w += gsl_matrix_get (idata->n, i, i); - w /= idata->n->size1; + for (i = 0; i < idata->mm.n->size1; ++i) + w += gsl_matrix_get (idata->mm.n, i, i); + w /= idata->mm.n->size1; xsq = w - 1 - (2 * factor->n_vars + 5) / 6.0; xsq *= -log (idata->detR); @@ -2210,10 +2295,11 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) } show_correlation_matrix (factor, idata); - covariance_destroy (cov); + if (idata->cvm) + covariance_destroy (idata->cvm); { - gsl_matrix *am = matrix_dup (analysis_matrix); + gsl_matrix *am = matrix_dup (idata->analysis_matrix); gsl_eigen_symmv_workspace *workspace = gsl_eigen_symmv_alloc (factor->n_vars); gsl_eigen_symmv (am, idata->eval, idata->evec, workspace); @@ -2256,11 +2342,11 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) if ( factor->extraction == EXTRACTION_PAF) { gsl_vector *diff = gsl_vector_alloc (idata->msr->size); - struct smr_workspace *ws = ws_create (analysis_matrix); + struct smr_workspace *ws = ws_create (idata->analysis_matrix); for (i = 0 ; i < factor->n_vars ; ++i) { - double r2 = squared_multiple_correlation (analysis_matrix, i, ws); + double r2 = squared_multiple_correlation (idata->analysis_matrix, i, ws); gsl_vector_set (idata->msr, i, r2); } @@ -2273,7 +2359,7 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) double min, max; gsl_vector_memcpy (diff, idata->msr); - iterate_factor_matrix (analysis_matrix, idata->msr, factor_matrix, fmw); + iterate_factor_matrix (idata->analysis_matrix, idata->msr, factor_matrix, fmw); gsl_vector_sub (diff, idata->msr); @@ -2296,7 +2382,7 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) gsl_vector_memcpy (extracted_communalities, initial_communalities); - iterate_factor_matrix (analysis_matrix, extracted_communalities, factor_matrix, fmw); + iterate_factor_matrix (idata->analysis_matrix, extracted_communalities, factor_matrix, fmw); extracted_eigenvalues = idata->eval; @@ -2340,7 +2426,8 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) { show_factor_matrix (factor, idata, (factor->rotation == ROT_PROMAX) ? _("Structure Matrix") : - (factor->extraction == EXTRACTION_PC ? _("Rotated Component Matrix") : _("Rotated Factor Matrix")), + (factor->extraction == EXTRACTION_PC ? _("Rotated Component Matrix") : + _("Rotated Factor Matrix")), rotated_factors); gsl_matrix_free (rotated_factors); @@ -2359,10 +2446,7 @@ do_factor (const struct cmd_factor *factor, struct casereader *r) } finish: - - idata_free (idata); - - casereader_destroy (r); + return; } -- 2.30.2