Merge pull request #316 from CartoDB/quantiles-bins-updates
Updates quantile bins to use native PostgreSQL precentile functions
This commit is contained in:
@@ -1,7 +1,7 @@
|
|||||||
# cartodb/Makefile
|
# cartodb/Makefile
|
||||||
|
|
||||||
EXTENSION = cartodb
|
EXTENSION = cartodb
|
||||||
EXTVERSION = 0.21.0
|
EXTVERSION = 0.22.0
|
||||||
|
|
||||||
SED = sed
|
SED = sed
|
||||||
AWK = awk
|
AWK = awk
|
||||||
@@ -85,6 +85,7 @@ UPGRADABLE = \
|
|||||||
0.19.2 \
|
0.19.2 \
|
||||||
0.20.0 \
|
0.20.0 \
|
||||||
0.21.0 \
|
0.21.0 \
|
||||||
|
0.22.0 \
|
||||||
$(EXTVERSION)dev \
|
$(EXTVERSION)dev \
|
||||||
$(EXTVERSION)next \
|
$(EXTVERSION)next \
|
||||||
$(END)
|
$(END)
|
||||||
|
|||||||
@@ -1,5 +1,6 @@
|
|||||||
0.22.0 (yyyy-mm-dd)
|
0.22.0 (yyyy-mm-dd)
|
||||||
* Fix: allow older ogr2ogr to work in -append mode (#319)
|
* Fix: allow older ogr2ogr to work in -append mode (#319)
|
||||||
|
* Refactors CDB_QuantileBins to rely on PostgreSQL function `percentile_disc` #316
|
||||||
|
|
||||||
0.21.0 (2018-02-15)
|
0.21.0 (2018-02-15)
|
||||||
* Add optional parameter to limit the number of cells in grid-generation functions #322
|
* Add optional parameter to limit the number of cells in grid-generation functions #322
|
||||||
|
|||||||
@@ -7,37 +7,12 @@
|
|||||||
-- @param breaks The number of bins you want to find.
|
-- @param breaks The number of bins you want to find.
|
||||||
--
|
--
|
||||||
--
|
--
|
||||||
CREATE OR REPLACE FUNCTION CDB_QuantileBins ( in_array NUMERIC[], breaks INT) RETURNS NUMERIC[] as $$
|
CREATE OR REPLACE FUNCTION CDB_QuantileBins(in_array numeric[], breaks int)
|
||||||
DECLARE
|
RETURNS numeric[]
|
||||||
element_count INT4;
|
AS $$
|
||||||
break_size numeric;
|
SELECT
|
||||||
tmp_val numeric;
|
percentile_disc(Array(SELECT generate_series(1, breaks) / breaks::numeric))
|
||||||
i INT := 1;
|
WITHIN GROUP (ORDER BY x ASC) AS p
|
||||||
reply numeric[];
|
FROM
|
||||||
BEGIN
|
unnest(in_array) AS x;
|
||||||
-- sort our values
|
$$ language SQL IMMUTABLE STRICT PARALLEL SAFE;
|
||||||
SELECT array_agg(e) INTO in_array FROM (SELECT unnest(in_array) e ORDER BY e ASC) x;
|
|
||||||
-- get the total size of our data
|
|
||||||
element_count := array_length(in_array, 1);
|
|
||||||
break_size := element_count::numeric / breaks;
|
|
||||||
-- slice our bread
|
|
||||||
LOOP
|
|
||||||
IF i < breaks THEN
|
|
||||||
IF break_size * i % 1 > 0 THEN
|
|
||||||
SELECT e INTO tmp_val FROM ( SELECT unnest(in_array) e LIMIT 1 OFFSET ceil(break_size * i) - 1) x;
|
|
||||||
ELSE
|
|
||||||
SELECT avg(e) INTO tmp_val FROM ( SELECT unnest(in_array) e LIMIT 2 OFFSET ceil(break_size * i) - 1 ) x;
|
|
||||||
END IF;
|
|
||||||
ELSIF i = breaks THEN
|
|
||||||
-- select the last value
|
|
||||||
SELECT max(e) INTO tmp_val FROM ( SELECT unnest(in_array) e ) x;
|
|
||||||
ELSE
|
|
||||||
EXIT;
|
|
||||||
END IF;
|
|
||||||
|
|
||||||
reply = array_append(reply, tmp_val);
|
|
||||||
i := i+1;
|
|
||||||
END LOOP;
|
|
||||||
RETURN reply;
|
|
||||||
END;
|
|
||||||
$$ language plpgsql IMMUTABLE STRICT PARALLEL SAFE;
|
|
||||||
|
|||||||
@@ -1,11 +1,23 @@
|
|||||||
WITH data AS (
|
WITH data AS (
|
||||||
SELECT array_agg(x::numeric) s FROM generate_series(1,300) x
|
SELECT Array[0.99, 1.0, 1.01,
|
||||||
WHERE x % 5 != 0 AND x % 7 != 0
|
4.99, 5.01,
|
||||||
)
|
10.01, 10.01,
|
||||||
SELECT unnest(CDB_JenksBins(s, 7)) FROM data;
|
15.01, 14.99,
|
||||||
|
20.1, 19.9]::numeric[] AS s
|
||||||
|
)
|
||||||
|
-- expectation is: 1, 5, 10, 15, 20
|
||||||
|
-- TODO: fix cdb_jenksbins to match ^^
|
||||||
|
SELECT round(unnest(CDB_JenksBins(s, 5))) FROM data;
|
||||||
|
|
||||||
WITH data_nulls AS (
|
WITH data_nulls AS (
|
||||||
SELECT array_agg(CASE WHEN x % 2 != 0 THEN x ELSE NULL END::numeric) s FROM generate_series(1,300) x
|
SELECT Array[0.99, 1.0, 1.01,
|
||||||
WHERE x % 5 != 0 AND x % 7 != 0
|
4.99, 5.01,
|
||||||
)
|
null, null,
|
||||||
SELECT unnest(CDB_JenksBins(s, 7)) FROM data_nulls;
|
10.01, 10.01,
|
||||||
|
15.01, 14.99,
|
||||||
|
null, null,
|
||||||
|
20.1, 19.9]::numeric[] AS s
|
||||||
|
)
|
||||||
|
-- expectation is: 1, 5, 10, 15, 20
|
||||||
|
-- TODO: fix cdb_jenksbins to match ^^
|
||||||
|
SELECT round(unnest(CDB_JenksBins(s, 5))) FROM data_nulls;
|
||||||
|
|||||||
@@ -1,14 +1,10 @@
|
|||||||
43
|
1
|
||||||
86
|
5
|
||||||
129
|
10
|
||||||
172
|
20
|
||||||
213
|
20
|
||||||
257
|
1
|
||||||
299
|
5
|
||||||
37
|
10
|
||||||
51
|
20
|
||||||
97
|
|
||||||
157
|
|
||||||
213
|
|
||||||
241
|
|
||||||
|
|
||||||
|
|||||||
@@ -1,11 +1,17 @@
|
|||||||
WITH data AS (
|
WITH data AS (
|
||||||
SELECT array_agg(x::numeric) s FROM generate_series(1,100) x
|
SELECT array_agg(x::numeric) AS s
|
||||||
WHERE x % 5 != 0 AND x % 7 != 0
|
FROM generate_series(0, 99) AS x
|
||||||
)
|
)
|
||||||
SELECT unnest(CDB_QuantileBins(s, 7)) FROM data;
|
SELECT unnest(CDB_QuantileBins(s, 10))
|
||||||
|
FROM data;
|
||||||
|
|
||||||
WITH data_nulls AS (
|
WITH data_nulls AS (
|
||||||
SELECT array_agg(CASE WHEN x % 2 != 0 THEN x ELSE NULL END::numeric) s FROM generate_series(1,100) x
|
SELECT array_agg(x::numeric) AS s
|
||||||
WHERE x % 5 != 0 AND x % 7 != 0
|
FROM (
|
||||||
|
SELECT x FROM generate_series(0, 99) AS x
|
||||||
|
UNION ALL
|
||||||
|
SELECT null AS x FROM generate_series(1, 10) AS x
|
||||||
|
) _wrap
|
||||||
)
|
)
|
||||||
SELECT unnest(CDB_QuantileBins(s, 7)) FROM data_nulls;
|
SELECT unnest(CDB_QuantileBins(s, 10))
|
||||||
|
FROM data_nulls;
|
||||||
|
|||||||
@@ -1,14 +1,20 @@
|
|||||||
13
|
9
|
||||||
|
19
|
||||||
29
|
29
|
||||||
43
|
39
|
||||||
57
|
49
|
||||||
71
|
59
|
||||||
86
|
69
|
||||||
|
79
|
||||||
|
89
|
||||||
99
|
99
|
||||||
|
9
|
||||||
|
19
|
||||||
29
|
29
|
||||||
57
|
39
|
||||||
87
|
49
|
||||||
|
59
|
||||||
|
69
|
||||||
|
79
|
||||||
|
89
|
||||||
99
|
99
|
||||||
|
|||||||
Reference in New Issue
Block a user