/* PSPP - a program for statistical analysis.
- Copyright (C) 2005 Free Software Foundation, Inc. Written by Jason H. Stover.
+ Copyright (C) 2005, 2011 Free Software Foundation, Inc. Written by Jason H. Stover.
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
#ifndef LINREG_H
#define LINREG_H
-#include <stdbool.h>
+
#include <gsl/gsl_math.h>
-#include <gsl/gsl_vector.h>
#include <gsl/gsl_matrix.h>
-#include <src/math/coefficient.h>
-#include <math/covariance-matrix.h>
+#include <gsl/gsl_vector.h>
+#include <stdbool.h>
enum
{
- PSPP_LINREG_CONDITIONAL_INVERSE,
- PSPP_LINREG_QR,
- PSPP_LINREG_SWEEP,
+ LINREG_CONDITIONAL_INVERSE,
+ LINREG_QR,
+ LINREG_SWEEP,
};
*/
-struct pspp_linreg_cache_struct
+struct linreg_struct
{
- int n_obs; /* Number of observations. */
+ double n_obs; /* Number of observations. */
int n_indeps; /* Number of independent variables. */
int n_coeffs; /* The intercept is not considered a
coefficient here. */
const struct variable *depvar;
const struct variable **indep_vars;
- gsl_vector *residuals;
- struct pspp_coeff **coeff;
+ double *coeff;
double intercept;
int method; /* Method to use to estimate parameters. */
/*
dfe, but since it is the best unbiased
estimate of the population variance, it
has its own entry here. */
- gsl_vector *ssx; /* Centered sums of squares for independent
- variables, i.e. \sum (x[i] - mean(x))^2. */
- double ssy; /* Centered sums of squares for dependent
- variable.
- */
/*
Covariance matrix of the parameter estimates.
*/
double dfe;
double dfm;
- /*
- 'Hat' or Hessian matrix, i.e. (X'X)^{-1}, where X is our
- design matrix.
- */
- gsl_matrix *hat;
-
- double (*predict) (const struct variable **, const union value **,
- const void *, int);
- double (*residual) (const struct variable **,
- const union value **,
- const union value *, const void *, int);
- /*
- Returns pointers to the variables used in the model.
- */
- int (*get_vars) (const void *, const struct variable **);
- struct variable *resid;
struct variable *pred;
-
+ struct variable *resid;
+ int dependent_column; /* Column containing the dependent variable. Defaults to last column. */
};
-typedef struct pspp_linreg_cache_struct pspp_linreg_cache;
+typedef struct linreg_struct linreg;
-/*
- Allocate a pspp_linreg_cache and return a pointer
- to it. n is the number of cases, p is the number of
- independent variables.
- */
-pspp_linreg_cache *pspp_linreg_cache_alloc (const struct variable *, const struct variable **,
- size_t, size_t);
+linreg *linreg_alloc (const struct variable *, const struct variable **,
+ double, size_t);
-bool pspp_linreg_cache_free (void *);
+bool linreg_free (void *);
/*
Fit the linear model via least squares. All pointers passed to pspp_linreg
are assumed to be allocated to the correct size and initialized to the
values as indicated by opts.
*/
-int
-pspp_linreg (const gsl_vector *, const struct design_matrix *,
- const pspp_linreg_opts *, pspp_linreg_cache *);
+void
+linreg_fit (const gsl_matrix *, linreg *);
double
-pspp_linreg_predict (const struct variable **, const union value **,
- const void *, int);
+linreg_predict (const linreg *, const double *, size_t);
double
-pspp_linreg_residual (const struct variable **, const union value **,
- const union value *, const void *, int);
-/*
- All variables used in the model.
- */
-int pspp_linreg_get_vars (const void *, const struct variable **);
+linreg_residual (const linreg *, double, const double *, size_t);
+const struct variable ** linreg_get_vars (const linreg *);
-struct pspp_coeff *pspp_linreg_get_coeff (const pspp_linreg_cache
- *,
- const struct variable
- *,
- const union value *);
/*
Return or set the standard deviation of the independent variable.
*/
-double pspp_linreg_get_indep_variable_sd (pspp_linreg_cache *, const struct variable *);
-void pspp_linreg_set_indep_variable_sd (pspp_linreg_cache *, const struct variable *, double);
+double linreg_get_indep_variable_sd (linreg *, size_t);
+void linreg_set_indep_variable_sd (linreg *, size_t, double);
/*
Mean of the independent variable.
*/
-double pspp_linreg_get_indep_variable_mean (pspp_linreg_cache *, const struct variable *);
-void pspp_linreg_set_indep_variable_mean (pspp_linreg_cache *, const struct variable *, double);
-
-/*
- Regression using only the covariance matrix.
- */
-void pspp_linreg_with_cov (const struct covariance_matrix *, pspp_linreg_cache *);
-double pspp_linreg_mse (const pspp_linreg_cache *);
+double linreg_get_indep_variable_mean (linreg *, size_t);
+void linreg_set_indep_variable_mean (linreg *, size_t, double);
+
+double linreg_mse (const linreg *);
+
+double linreg_intercept (const linreg *);
+
+gsl_matrix * linreg_cov (const linreg *);
+double linreg_coeff (const linreg *, size_t);
+const struct variable * linreg_indep_var (const linreg *, size_t);
+size_t linreg_n_coeffs (const linreg *);
+double linreg_n_obs (const linreg *);
+double linreg_sse (const linreg *);
+double linreg_ssreg (const linreg *);
+double linreg_dfmodel (const linreg *);
+double linreg_sst (const linreg *);
+void linreg_set_depvar_mean (linreg *, double);
+double linreg_get_depvar_mean (linreg *);
#endif