1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2005 Free Software Foundation, Inc. Written by Jason H. Stover.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
20 #include <gsl/gsl_math.h>
21 #include <gsl/gsl_vector.h>
22 #include <gsl/gsl_matrix.h>
26 LINREG_CONDITIONAL_INVERSE,
34 Options describing what special values should be computed.
36 struct pspp_linreg_opts_struct
38 int get_depvar_mean_std;
39 int *get_indep_mean_std; /* Array of booleans
41 independent variables need
42 their means and standard
43 deviations computed within
44 pspp_linreg. This array
46 n_indeps. If element i is
49 variance of indpendent
50 variable i. If element i
51 is 0, it will not compute
53 deviation, and assume the
55 cache->indep_mean[i] is
57 cache->indep_std[i] is the
58 sample standard deviation. */
60 typedef struct pspp_linreg_opts_struct pspp_linreg_opts;
64 Find the least-squares estimate of b for the linear model:
68 where Y is an n-by-1 column vector, X is an n-by-p matrix of
69 independent variables, b is a p-by-1 vector of regression coefficients,
70 and Z is an n-by-1 normally-distributed random vector with independent
71 identically distributed components with mean 0.
73 This estimate is found via the sweep operator or singular-value
74 decomposition with gsl.
79 1. Matrix Computations, third edition. GH Golub and CF Van Loan.
80 The Johns Hopkins University Press. 1996. ISBN 0-8018-5414-8.
82 2. Numerical Analysis for Statisticians. K Lange. Springer. 1999.
85 3. Numerical Linear Algebra for Applications in Statistics. JE Gentle.
86 Springer. 1998. ISBN 0-387-98542-5.
92 double n_obs; /* Number of observations. */
93 int n_indeps; /* Number of independent variables. */
94 int n_coeffs; /* The intercept is not considered a
98 Pointers to the variables.
100 const struct variable *depvar;
101 const struct variable **indep_vars;
105 int method; /* Method to use to estimate parameters. */
107 Means and standard deviations of the variables.
108 If these pointers are null when pspp_linreg() is
109 called, pspp_linreg() will compute their values.
111 Entry i of indep_means is the mean of independent
112 variable i, whose observations are stored in the ith
113 column of the design matrix.
117 gsl_vector *indep_means;
118 gsl_vector *indep_std;
123 double ssm; /* Sums of squares for the overall model. */
124 gsl_vector *ss_indeps; /* Sums of squares from each
125 independent variable. */
126 double sst; /* Sum of squares total. */
127 double sse; /* Sum of squares error. */
128 double mse; /* Mean squared error. This is just sse /
129 dfe, but since it is the best unbiased
130 estimate of the population variance, it
131 has its own entry here. */
133 Covariance matrix of the parameter estimates.
143 struct variable *pred;
144 struct variable *resid;
145 int dependent_column; /* Column containing the dependent variable. Defaults to last column. */
148 typedef struct linreg_struct linreg;
152 linreg *linreg_alloc (const struct variable *, const struct variable **,
155 bool linreg_free (void *);
158 Fit the linear model via least squares. All pointers passed to pspp_linreg
159 are assumed to be allocated to the correct size and initialized to the
160 values as indicated by opts.
163 linreg_fit (const gsl_matrix *, linreg *);
166 linreg_predict (const linreg *, const double *, size_t);
168 linreg_residual (const linreg *, double, const double *, size_t);
169 const struct variable ** linreg_get_vars (const linreg *);
172 Return or set the standard deviation of the independent variable.
174 double linreg_get_indep_variable_sd (linreg *, size_t);
175 void linreg_set_indep_variable_sd (linreg *, size_t, double);
177 Mean of the independent variable.
179 double linreg_get_indep_variable_mean (linreg *, size_t);
180 void linreg_set_indep_variable_mean (linreg *, size_t, double);
182 double linreg_mse (const linreg *);
184 double linreg_intercept (const linreg *);
186 gsl_matrix * linreg_cov (const linreg *);
187 double linreg_coeff (const linreg *, size_t);
188 const struct variable * linreg_indep_var (const linreg *, size_t);
189 size_t linreg_n_coeffs (const linreg *);
190 double linreg_n_obs (const linreg *);
191 double linreg_sse (const linreg *);
192 double linreg_ssreg (const linreg *);
193 double linreg_dfmodel (const linreg *);
194 double linreg_sst (const linreg *);
195 void linreg_set_depvar_mean (linreg *, double);
196 double linreg_get_depvar_mean (linreg *);