Merge pull request #80 from CartoDB/sariogonfer-master

updates quantile bins algorithm

scripts-available/CDB_QuantileBins.sql

@@ -15,18 +15,29 @@ DECLARE
     i INT := 1; 
     reply numeric[]; 
 BEGIN 
-    -- get our unique values
+    -- sort our values
-    SELECT array_agg(e) INTO in_array FROM (SELECT unnest(in_array) e GROUP BY e ORDER BY e ASC) x;
+    SELECT array_agg(e) INTO in_array FROM (SELECT unnest(in_array) e ORDER BY e ASC) x;
-    -- get the total size of our row
+    -- get the total size of our data
-    element_count := array_upper(in_array, 1) - array_lower(in_array, 1); 
+    element_count := array_length(in_array, 1); 
     break_size :=  element_count::numeric / breaks;
     -- slice our bread
     LOOP  
-        IF i > breaks THEN  EXIT;  END IF;  
+        IF i < breaks THEN
-        SELECT e INTO tmp_val FROM ( SELECT unnest(in_array) e LIMIT 1 OFFSET round(break_size * i)) x;
+            IF break_size * i % 1 > 0 THEN
+                SELECT e INTO tmp_val FROM ( SELECT unnest(in_array) e LIMIT 1 OFFSET ceil(break_size * i) - 1) x;
+            ELSE
+                SELECT avg(e) INTO tmp_val FROM ( SELECT unnest(in_array) e LIMIT 2 OFFSET ceil(break_size * i) - 1 ) x;
+            END IF;
+        ELSIF i = breaks THEN
+            -- select the last value
+            SELECT max(e) INTO tmp_val FROM ( SELECT unnest(in_array) e ) x;
+        ELSE
+            EXIT;
+        END IF;
+
         reply = array_append(reply, tmp_val);
-        i := i+1; 
+        i := i+1;
-    END LOOP; 
+    END LOOP;
-    RETURN reply; 
+    RETURN reply;
 END; 
-$$ language plpgsql IMMUTABLE;
+$$ language plpgsql IMMUTABLE;

test/CDB_QuantileBinsTest_expect

@@ -1,7 +1,7 @@
-16
+13
 29
 43
 57
 71
-83
+86
 99