From db323f3e13574cf700c62e5a2001d669f1152960 Mon Sep 17 00:00:00 2001
From: Andy Eschbacher <eschbacher@cartodb.com>
Date: Tue, 25 Aug 2015 22:58:31 -0400
Subject: [PATCH] adding skewness

---
 scripts-available/CDB_Stats.sql | 39 +++++++++++++++++++++++----------
 1 file changed, 28 insertions(+), 11 deletions(-)

diff --git a/scripts-available/CDB_Stats.sql b/scripts-available/CDB_Stats.sql
index f0a47ba..b9b765b 100644
--- a/scripts-available/CDB_Stats.sql
+++ b/scripts-available/CDB_Stats.sql
@@ -1,18 +1,15 @@
 --
--- Calculate the Pearson kurtosis of the input data
+-- Calculate basic statistics of a given dataset
 --
--- @param in_array A numeric array of numbers to determine the best
---                   to determine the bin boundary
+-- @param in_array A numeric array of numbers
 --
--- @param breaks The number of bins you want to find.
---  
---
--- Returns: upper edges of bins
+-- Returns: statistical quantity chosen
 -- 
 --
 
+-- Calculate Pearson's moment coefficient of kurtosis
 CREATE OR REPLACE FUNCTION CDB_Kurtosis ( in_array NUMERIC[] ) RETURNS NUMERIC as $$
-DECLARE 
+DECLARE
     a numeric;
     c numeric;
     s numeric;
@@ -22,12 +19,32 @@ BEGIN
 
     RAISE NOTICE 'avg: %, cnt: %, std: %', a, c, s;
 
-    EXECUTE '
-        SELECT sum(power($1 - e,4)) / ( $2 * power($3, 4))
-        FROM (SELECT unnest($4) e ) x'
+    EXECUTE 'SELECT sum(power($1 - e, 4)) / ( $2 * power($3, 4))
+             FROM (SELECT unnest($4) e ) x'
     INTO k
     USING a, c, s, in_array;
 
     RETURN k;
 END;
 $$ language plpgsql IMMUTABLE;
+
+-- Calculate Pearson's moment coefficient of skewness
+CREATE OR REPLACE FUNCTION CDB_Skewness ( in_array NUMERIC[] ) RETURNS NUMERIC as $$
+DECLARE
+    a numeric;
+    c numeric;
+    s numeric;
+    sk numeric;
+BEGIN
+    SELECT AVG(e), COUNT(e)::numeric, stddev(e) INTO a, c, s FROM ( SELECT unnest(in_array) e ) x;
+
+    RAISE NOTICE 'avg: %, cnt: %, std: %', a, c, s;
+
+    EXECUTE 'SELECT sum(power($1 - e, 3)) / ( $2 * power($3, 3))
+             FROM (SELECT unnest($4) e ) x'
+    INTO sk
+    USING a, c, s, in_array;
+
+    RETURN sk;
+END;
+$$ language plpgsql IMMUTABLE;