X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.h;h=c5919bc455e883e131bb2aa637b1165d0bf36ea2;hb=e1335ea67114822554e0d51bd5e0b49401b17dc4;hp=02fef13a303d17d717ef07df8b54c58a5302a2cc;hpb=c91d15f0e87c2667ae4b39c236850ef6669a6b51;p=pspp diff --git a/src/math/categoricals.h b/src/math/categoricals.h index 02fef13a30..c5919bc455 100644 --- a/src/math/categoricals.h +++ b/src/math/categoricals.h @@ -1,5 +1,5 @@ /* PSPP - a program for statistical analysis. - Copyright (C) 2009 Free Software Foundation, Inc. + Copyright (C) 2009, 2010, 2011, 2012 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 @@ -19,26 +19,18 @@ #define _CATEGORICALS__ #include -#include +#include "data/missing-values.h" struct categoricals; struct variable; struct ccase; +struct interaction; union value ; -typedef void update_func (void *user_data, - const struct variable *wv, - const struct variable *catvar, - const struct ccase *c, - void *aux1, void *aux2); - -typedef void *user_data_create_func (void *aux1, void *aux2); - -struct categoricals *categoricals_create (const struct variable *const *v, size_t n_vars, - const struct variable *wv, enum mv_class exclude, - user_data_create_func *udf, - update_func *update, void *aux1, void *aux2); +struct categoricals *categoricals_create (struct interaction *const*, size_t n_int, + const struct variable *wv, enum mv_class dep_excl, + enum mv_class fctr_excl); void categoricals_destroy (struct categoricals *); @@ -48,9 +40,14 @@ void categoricals_update (struct categoricals *cat, const struct ccase *c); /* Return the number of categories (distinct values) for variable N */ size_t categoricals_n_count (const struct categoricals *cat, size_t n); +size_t categoricals_df (const struct categoricals *cat, size_t n); /* Return the total number of categories */ -size_t categoricals_total (const struct categoricals *cat); +size_t categoricals_n_total (const struct categoricals *cat); + +/* Return the total degrees of freedom */ +size_t categoricals_df_total (const struct categoricals *cat); + /* Return the total number of variables which participated in these categoricals. @@ -60,23 +57,72 @@ size_t categoricals_total (const struct categoricals *cat); */ size_t categoricals_get_n_variables (const struct categoricals *cat); -void categoricals_done (struct categoricals *cat); +bool categoricals_is_complete (const struct categoricals *cat); -const struct variable * categoricals_get_variable_by_subscript (const struct categoricals *cat, int subscript); -const union value * categoricals_get_value_by_subscript (const struct categoricals *cat, int subscript); +/* + Must be called (once) before any call to the *_by_subscript or *_by_category + functions, but AFTER any calls to categoricals_update. + If this function returns false, then no calls to _by_subscript or *_by_category + are allowed. +*/ +void categoricals_done (const struct categoricals *cat); + +bool categoricals_sane (const struct categoricals *cat); + +/* + The *_by_subscript functions use the short map. + Their intended use is by covariance matrix routines, where normally 1 less than + the total number of distinct values of each categorical variable should + be considered. + */ double categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscript); +const struct interaction *categoricals_get_interaction_by_subscript (const struct categoricals *cat, int subscript); double categoricals_get_sum_by_subscript (const struct categoricals *cat, int subscript); -double categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscript, - const struct ccase *c); +/* Returns unity if the value in case C at SUBSCRIPT is equal to the category + for that subscript */ +double +categoricals_get_dummy_code_for_case (const struct categoricals *cat, int subscript, + const struct ccase *c); + +/* Returns unity if the value in case C at SUBSCRIPT is equal to the category + for that subscript. + Else if it is the last category, return -1. + Otherwise return 0. + */ +double +categoricals_get_effects_code_for_case (const struct categoricals *cat, int subscript, + const struct ccase *c); + + +/* These use the long map. Useful for descriptive statistics. */ + + +const struct ccase * +categoricals_get_case_by_category_real (const struct categoricals *cat, int iact, int n); + +void * +categoricals_get_user_data_by_category_real (const struct categoricals *cat, int iact, int n); + + +void * categoricals_get_user_data_by_category (const struct categoricals *cat, int category); + +const struct ccase * categoricals_get_case_by_category (const struct categoricals *cat, int subscript); -void * categoricals_get_user_data_by_subscript (const struct categoricals *cat, int subscript); +struct payload +{ + void* (*create) (const void *aux1, void *aux2); + void (*update) (const void *aux1, void *aux2, void *user_data, const struct ccase *, double weight); + void (*calculate) (const void *aux1, void *aux2, void *user_data); + void (*destroy) (const void *aux1, void *aux2, void *user_data); +}; +void categoricals_set_payload (struct categoricals *cats, const struct payload *p, const void *aux1, void *aux2); #endif