Increase default percentiles agg compression to 200

CHANGELOG.md

@@ -21,6 +21,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Onboarding new maven snapshots publishing to s3 ([#19619](https://github.com/opensearch-project/OpenSearch/pull/19619))
 - Remove MultiCollectorWrapper and use MultiCollector in Lucene instead ([#19595](https://github.com/opensearch-project/OpenSearch/pull/19595))
 - Change implementation for `percentiles` aggregation for latency improvement ([#19648](https://github.com/opensearch-project/OpenSearch/pull/19648))
+- Change default `percentiles` aggregation compression to 200 for accuracy improvement ([#18459](https://github.com/opensearch-project/OpenSearch/pull/18459))
 
 ### Fixed
 - Fix Allocation and Rebalance Constraints of WeightFunction are incorrectly reset ([#19012](https://github.com/opensearch-project/OpenSearch/pull/19012))

server/src/main/java/org/opensearch/search/aggregations/metrics/PercentilesConfig.java

@@ -129,7 +129,7 @@ public int hashCode() {
      * @opensearch.internal
      */
     public static class TDigest extends PercentilesConfig {
-        static final double DEFAULT_COMPRESSION = 100.0;
+        static final double DEFAULT_COMPRESSION = 200.0;
         private double compression;
 
         public TDigest() {

server/src/test/java/org/opensearch/search/builder/SearchSourceBuilderTests.java

@@ -137,7 +137,7 @@ public void testSerializationWithPercentilesQueryObject() throws IOException {
             + "\"percents\":[1.0,5.0,25.0,50.0,75.0,95.0,99.0],"
             + "\"keyed\":true,"
             + "\"tdigest\":{"
-            + "\"compression\":100.0"
+            + "\"compression\":200.0"
             + "}"
             + "}"
             + "}"