Implemented calculation of percentiles and Tukey hinges

[pspp-builds.git] / src / factor_stats.h

diff --git a/src/factor_stats.h b/src/factor_stats.h
index 67fd5f520ebcbf52ff54c5293959b38da312aa95..f6394a8a95aa1a8358b90e1812edb32d521bfc12 100644 (file)
--- a/src/factor_stats.h
+++ b/src/factor_stats.h
@@ -25,60 +25,137 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
 /* FIXME: These things should probably be amalgamated with the 
    group_statistics struct */
 
+#include "hash.h"
+#include "val.h"
+#include <gsl/gsl_histogram.h>
+#include "subclist.h"
+#include "percentiles.h"
+
+struct moments1;
 
 struct metrics
 {
   double n;
-  
-  double ssq;
-  
-  double sum;
 
+  double n_missing;
+  
   double min;
 
   double max;
 
   double mean;
   
-  double stderr;
+  double se_mean;
 
   double var;
 
   double stddev;
+
+  struct moments1 *moments;
+
+  gsl_histogram *histogram;
+
+  double skewness;
+  double kurtosis;
+
+  double trimmed_mean;
+
+  /* A hash of data for this factor. */
+  struct hsh_table *ordered_data;
+
+  /* A Pointer to this hash table AFTER it has been SORTED and crunched */
+  struct weighted_value **wvp;
+
+  /* The number of values in the above array
+     (if all the weights are 1, then this will
+     be the same as n) */
+  int n_data;
+
+  /* Percentile stuff */
+
+  /* A hash of struct percentiles */
+  struct hsh_table *ptile_hash;
+
+  /* Algorithm to be used for calculating percentiles */
+  enum pc_alg ptile_alg;
+
+  /* Tukey's Hinges */
+  double hinges[3];
+
 };
 
 
 
-struct factor_statistics {
 
-  /* The value of the independent variable for this factor */
-  const union value *id;
+void metrics_precalc(struct metrics *m);
 
-  /* An array of metrics indexed by dependent variable */
-  struct metrics *stats;
+void metrics_calc(struct metrics *m, const union value *f, double weight, 
+                 int case_no);
 
+void metrics_postcalc(struct metrics *m);
+
+
+/* Linked list of case nos */
+struct case_node
+{
+  int num;
+  struct case_node *next;
 };
 
+struct weighted_value 
+{
+  union value v;
+
+  /* The weight */
+  double w;
 
+  /* The cumulative weight */
+  double cc; 
+
+  /* The rank */
+  double rank;
+
+  /* Linked list of cases nos which have this value */
+  struct case_node *case_nos;
+  
+};
 
-void metrics_precalc(struct metrics *fs);
 
-void metrics_calc(struct metrics *fs, double x, double weight);
+struct weighted_value *weighted_value_create(void);
 
-void metrics_postcalc(struct metrics *fs);
+void weighted_value_free(struct weighted_value *wv);
+
+
+
+struct factor_statistics {
+
+  /* The values of the independent variables */
+  union value id[2];
+
+  /* The an array stats for this factor, one for each dependent var */
+  struct metrics *m;
+
+};
 
 
+/* Create a factor statistics object with for N dependent vars
+   and ID as the value of the independent variable */
+struct factor_statistics * 
+create_factor_statistics (int n, union value *id0, union value *id1);
 
 
-/* These functions are necessary for creating hashes */
+void factor_statistics_free(struct factor_statistics *f);
 
-int compare_indep_values(const struct factor_statistics *f1, 
-                    const struct factor_statistics *f2, 
-                    int width);
 
-unsigned hash_indep_value(const struct factor_statistics *f, int width) ;
+/* Compare f0 and f1.
+   width is the width of the independent variable */
+int 
+factor_statistics_compare(const struct factor_statistics *f0,
+                         const struct factor_statistics *f1, int width);
 
-void  free_factor_stats(struct factor_statistics *f, int width );
+                             
 
+unsigned int 
+factor_statistics_hash(const struct factor_statistics *f, int width);
 
 #endif