New contrib module for statistical mathematics

Jan

contrib/statmath/Makefile

@@ -0,0 +1,58 @@
+#
+# $Header: /cvsroot/pgsql/contrib/statmath/Attic/Makefile,v 1.1 2000/07/04 14:36:03 wieck Exp $
+#
+
+TOPDIR=../..
+
+include ../Makefile.global
+
+NAME	= statmath
+
+PROGRAM	= 
+OBJS	= $(NAME).o
+DOCS	= README.$(NAME)
+SQLS	= $(NAME).sql
+BINS	=
+EXAMPLES=
+MODS	= $(NAME)$(DLSUFFIX)
+
+CFLAGS += -I. $(CFLAGS_SL)
+
+OTHER_CLEAN = $(SQLS)
+
+all: $(MODS) $(SQLS)
+
+%.sql: %.sql.in
+	$(SED) "s|MODULE_PATHNAME|$(CONTRIB_MODDIR)/$(NAME)$(DLSUFFIX)|" < $< > $@
+
+install: install_doc install_sql install_mod install_bin
+
+install_doc:
+	for inst_file in $(DOCS); do \
+		$(INSTALL) $(INSTL_LIB_OPTS) $$inst_file $(CONTRIB_DOCDIR); \
+	done
+
+install_sql:
+	for inst_file in $(SQLS); do \
+		$(INSTALL) $(INSTL_LIB_OPTS) $$inst_file $(CONTRIB_SQLDIR); \
+	done
+
+install_mod:
+	for inst_file in $(MODS); do \
+		$(INSTALL) $(INSTL_SHLIB_OPTS) $$inst_file $(CONTRIB_MODDIR); \
+	done
+
+install_bin:
+	for inst_file in $(BINS); do \
+		$(INSTALL) $(INSTL_EXE_OPTS) $$inst_file $(CONTRIB_BINDIR); \
+	done
+
+depend dep:
+	$(CC) -MM -MG $(CFLAGS) *.c > depend
+
+clean:
+	$(RM) *~ $(OBJS) $(MODS) $(PROGRAM) depend $(OTHER_CLEAN) core log
+
+ifeq (depend,$(wildcard depend))
+include depend
+endif

contrib/statmath/README.statmath

@@ -0,0 +1,39 @@
+Statistical aggregate functions for PostgreSQL.
+
+This module provides some aggregate functions for statistical 
+mathematics. A new datatype holding two double precision 
+floating point values is required by them. 
+
+The aggregates are:
+
+average(float8)         average value computed by
+
+                                   n
+                                  ___
+                        _    1    \
+                        x = ---    >  x
+                             n    /__  i
+                                  i=1
+
+
+variance(float8         variance computed by
+
+                                    n
+                                   ___
+                         2    1    \         _ 2
+                        s  = ---    >  (x  - x) 
+                             n-1   /__   i
+                                   i=1
+
+
+
+stddev(float8)          standard deviation computed by
+
+                                 _____
+                                /  2 
+                        s =    /  s
+                             \/
+
+
+-- 
+Jan Wieck <JanWieck@Yahoo.com>

contrib/statmath/statmath.c

@@ -0,0 +1,307 @@
+/* ----------
+ * Module statmath
+ *
+ *	statistical aggregates for average, variance and standard
+ *	deviation.
+ *
+ *	Jan Wieck
+ * ----------
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <math.h>
+#include <errno.h>
+
+#include "postgres.h"
+#include "utils/palloc.h"
+
+
+/* ----------
+ * Declarations
+ *
+ * statmath_stateval_in()	Input function for state transition variable
+ * statmath_stateval_out()	Output function for state transition variable
+ * statmath_collect()		State transition function to collect items
+ * statmath_average_fin()	Final function for average aggregate
+ * statmath_variance_fin()	Final function for variance aggregate
+ * statmath_stddev_fin()	Final function for deviation aggregate
+ * ----------
+ */
+
+float64 statmath_stateval_in(char *str);
+char *statmath_stateval_out(float64 sval);
+
+float64 statmath_collect(float64 sval, float64 item);
+float64 statmath_average_fin(float64 sval, int4 n);
+float64 statmath_variance_fin(float64 sval, int4 n);
+float64 statmath_stddev_fin(float64 sval, int4 n);
+
+
+/* ----------
+ * statmath_checkval -
+ *
+ *	Bounds checking for float8 values in Postgres
+ * ----------
+ */
+static void
+statmath_checkval(double val)
+{
+	if (fabs(val) > DBL_MAX)
+		elog(ERROR, "statmath: overflow");
+	if (val != 0.0 && fabs(val) < DBL_MIN)
+		elog(ERROR, "statmath: underflow");
+}
+
+
+/* ----------
+ * statmath_stateval_in -
+ *
+ *	Input function for the state transition value data type.
+ *	The input string are two float8's separated with a colon ':'.
+ * ----------
+ */
+float64
+statmath_stateval_in(char *str)
+{
+	float64		retval;
+	double		tmp;
+	char		*cp1, *cp2;
+
+	if (!str)
+		return (float64) NULL;
+
+	/*
+	 * Allocate space for the result
+	 */
+	retval = (float64) palloc(sizeof(float64data) * 2);
+
+	/*
+	 * Get the first number
+	 */
+	errno = 0;
+	tmp = strtod(str, &cp1);
+	if (*cp1 != ':' || errno == ERANGE)
+		elog(ERROR, "statmath: illegal input format '%s'", str);
+	statmath_checkval(tmp);
+	retval[0] = tmp;
+
+	/*
+	 * Get the second number
+	 */
+	tmp = strtod(++cp1, &cp2);
+	if (*cp2 != '\0' || errno == ERANGE)
+		elog(ERROR, "statmath: illegal input format '%s'", str);
+	statmath_checkval(tmp);
+	retval[1] = tmp;
+
+	/*
+	 * Return the internal binary format
+	 */
+	return retval;
+}
+
+
+/* ----------
+ * statmath_stateval_out -
+ *
+ *	Output function for the state transition value data type.
+ * ----------
+ */
+char *
+statmath_stateval_out(float64 sval)
+{
+	char		buf[1024];
+	double		v1, v2;
+
+	if (!sval)
+		return pstrdup("(null)");
+	
+	/*
+	 * Print the values in the external format and return
+	 * the result in allocated space
+	 */
+	v1 = sval[0];
+	v2 = sval[1];
+	sprintf(buf, "%.*g:%.*g", DBL_DIG, v1, DBL_DIG, v2);
+	return pstrdup(buf);
+}
+
+
+/* ----------
+ * statmath_collect -
+ *
+ *	State transition function to collect data for the variance
+ *	and standard deviation aggregates.
+ *	The state transition variable holds 2 float8 values. The
+ *	first is the sum of the items, the second the sum of the
+ *	item quadratic products.
+ * ----------
+ */
+float64
+statmath_collect(float64 sval, float64 item)
+{
+	float64		retval;
+	double		tmp;
+
+	if (!sval || !item)
+		return (float64) NULL;
+
+	/*
+	 * Allocate space for the result
+	 */
+	retval = (float64) palloc(sizeof(float64data) * 2);
+
+	/*
+	 * Compute the new values
+	 */
+	tmp = sval[0] + *item;
+	statmath_checkval(tmp);
+	retval[0] = tmp;
+
+	tmp = sval[1] + *item * *item;
+	statmath_checkval(tmp);
+	retval[1] = tmp;
+
+	/*
+	 * Return the result
+	 */
+	return retval;
+}
+
+
+/* ----------
+ * statmath_average_fin -
+ *
+ *	Final computation function for the average aggregate.
+ * ----------
+ */
+float64
+statmath_average_fin(float64 sum, int4 n)
+{
+	float64		retval;
+	double		tmp;
+
+	if (!sum)
+		return (float64) NULL;
+
+	/*
+	 * Allocate space for the result
+	 */
+	retval = (float64) palloc(sizeof(float64data));
+
+	/*
+	 * Avoid division by zero if no items collected
+	 */
+	if (n == 0)
+	{
+		*retval = 0.0;
+		return retval;
+	}
+
+	/*
+	 * Compute the average
+	 */
+	tmp = *sum / (double)n;
+	statmath_checkval(tmp);
+	*retval = tmp;
+
+	/*
+	 * Return the result
+	 */
+	return retval;
+}
+
+
+/* ----------
+ * statmath_variance_fin -
+ *
+ *	Final computation function for the variance aggregate
+ * ----------
+ */
+float64
+statmath_variance_fin(float64 sval, int4 n)
+{
+	float64		retval;
+	double		avg;
+	double		variance;
+
+	if (!sval)
+		return (float64) NULL;
+
+	/*
+	 * Allocate space for the result
+	 */
+	retval = (float64) palloc(sizeof(float64data));
+
+	/*
+	 * Avoid division by zero if less than 2 items collected
+	 */
+	if (n < 2)
+	{
+		*retval = 0.0;
+		return retval;
+	}
+
+	/*
+	 * Calculate the variance
+	 */
+	avg = sval[0] / (double)n;
+	variance = (sval[1] - sval[0] * avg) / ((double)n - 1.0);
+
+	statmath_checkval(variance);
+	*retval = variance;
+
+	/*
+	 * Return the result
+	 */
+	return retval;
+}
+
+
+/* ----------
+ * statmath_stateval_in -
+ *
+ *	Input function for the state transition value data type
+ * ----------
+ */
+float64
+statmath_stddev_fin(float64 sval, int4 n)
+{
+	float64		retval;
+	double		avg;
+	double		stddev;
+
+	if (!sval)
+		return (float64) NULL;
+
+	/*
+	 * Allocate space for the result
+	 */
+	retval = (float64) palloc(sizeof(float64data));
+
+	/*
+	 * Avoid division by zero if less than 2 items collected
+	 */
+	if (n < 2)
+	{
+		*retval = 0.0;
+		return retval;
+	}
+
+	/*
+	 * Calculate the standard deviation
+	 */
+	avg = sval[0] / (double)n;
+	stddev = sqrt((sval[1] - sval[0] * avg) / ((double)n - 1.0));
+
+	statmath_checkval(stddev);
+	*retval = stddev;
+
+	/*
+	 * Return the result
+	 */
+	return retval;
+}
+
+

contrib/statmath/statmath.sql.in

@@ -0,0 +1,88 @@
+-- statmath.sql
+--
+-- Install the statistical aggregates
+--
+
+--
+-- Create the new data type for the state transition variable
+--
+CREATE FUNCTION statmath_stateval_in(opaque)
+	RETURNS statmath_stateval
+	AS 'MODULE_PATHNAME'
+	LANGUAGE 'C';
+
+CREATE FUNCTION statmath_stateval_out(opaque)
+	RETURNS opaque
+	AS 'MODULE_PATHNAME'
+	LANGUAGE 'C';
+
+CREATE TYPE statmath_stateval (
+	internallength = 16,
+	input = statmath_stateval_in,
+	output = statmath_stateval_out,
+	alignment = double
+);
+
+--
+-- Create the statistic data collector used in the aggregates
+--
+CREATE FUNCTION statmath_collect(statmath_stateval, float8)
+	RETURNS statmath_stateval
+	AS 'MODULE_PATHNAME'
+	LANGUAGE 'C';
+
+--
+-- Create the final functions for the three aggregates
+--
+CREATE FUNCTION statmath_average_fin(float8, int4)
+	RETURNS float8
+	AS 'MODULE_PATHNAME'
+	LANGUAGE 'C';
+
+CREATE FUNCTION statmath_variance_fin(statmath_stateval, int4)
+	RETURNS float8
+	AS 'MODULE_PATHNAME'
+	LANGUAGE 'C';
+
+CREATE FUNCTION statmath_stddev_fin(statmath_stateval, int4)
+	RETURNS float8
+	AS 'MODULE_PATHNAME'
+	LANGUAGE 'C';
+
+--
+-- Create the aggregates themself
+--
+CREATE AGGREGATE average (
+	basetype = float8,
+	stype1 = float8,
+	stype2 = int4,
+	sfunc1 = float8pl,
+	sfunc2 = int4inc,
+	finalfunc = statmath_average_fin,
+	initcond1 = '0',
+	initcond2 = '0'
+);
+
+CREATE AGGREGATE variance (
+	basetype = float8,
+	stype1 = statmath_stateval,
+	stype2 = int4,
+	sfunc1 = statmath_collect,
+	sfunc2 = int4inc,
+	finalfunc = statmath_variance_fin,
+	initcond1 = '0:0',
+	initcond2 = '0'
+);
+
+CREATE AGGREGATE stddev (
+	basetype = float8,
+	stype1 = statmath_stateval,
+	stype2 = int4,
+	sfunc1 = statmath_collect,
+	sfunc2 = int4inc,
+	finalfunc = statmath_stddev_fin,
+	initcond1 = '0:0',
+	initcond2 = '0'
+);
+
+