[improvement](statistics)Set min max to NULL when collect stats with sample (#25593)

1. To avoid misleading of inaccurate min max stats, set the stats value to NULL while using sample to collect stats.
2. Fix NDV_SAMPLE_TEMPLATE typo, it shouldn't contain row count related contents.

fe/fe-core/src/main/java/org/apache/doris/statistics/BaseAnalysisTask.java

@@ -46,8 +46,7 @@ public abstract class BaseAnalysisTask {
 
     protected static final String NDV_MULTIPLY_THRESHOLD = "0.3";
 
-    protected static final String NDV_SAMPLE_TEMPLATE = "ROUND(COUNT(1) * ${scaleFactor}) AS row_count, "
-            + "case when NDV(`${colName}`)/count('${colName}') < "
+    protected static final String NDV_SAMPLE_TEMPLATE = "case when NDV(`${colName}`)/count('${colName}') < "
             + NDV_MULTIPLY_THRESHOLD
             + " then NDV(`${colName}`) "
             + "else NDV(`${colName}`) * ${scaleFactor} end AS ndv, "
@@ -239,6 +238,24 @@ public abstract class BaseAnalysisTask {
         return "COUNT(1) * " + column.getType().getSlotSize();
     }
 
+    // Min value is not accurate while sample, so set it to NULL to avoid optimizer generate bad plan.
+    protected String getMinFunction() {
+        if (tableSample == null) {
+            return "MIN(`${colName}`) ";
+        } else {
+            return "NULL ";
+        }
+    }
+
+    // Max value is not accurate while sample, so set it to NULL to avoid optimizer generate bad plan.
+    protected String getMaxFunction() {
+        if (tableSample == null) {
+            return "MAX(`${colName}`) ";
+        } else {
+            return "NULL ";
+        }
+    }
+
     protected TableSample getTableSample() {
         if (info.forceFull) {
             return null;

fe/fe-core/src/main/java/org/apache/doris/statistics/HMSAnalysisTask.java

@@ -61,10 +61,11 @@ public class HMSAnalysisTask extends BaseAnalysisTask {
             + "${idxId} AS idx_id, "
             + "'${colId}' AS col_id, "
             + "NULL AS part_id, "
+            + "ROUND(COUNT(1) * ${scaleFactor}) AS row_count, "
             + NDV_SAMPLE_TEMPLATE
             + "ROUND(SUM(CASE WHEN `${colName}` IS NULL THEN 1 ELSE 0 END) * ${scaleFactor}) AS null_count, "
-            + "MIN(`${colName}`) AS min, "
-            + "MAX(`${colName}`) AS max, "
+            + "${minFunction} AS min, "
+            + "${maxFunction} AS max, "
             + "${dataSizeFunction} * ${scaleFactor} AS data_size, "
             + "NOW() "
             + "FROM `${catalogName}`.`${dbName}`.`${tblName}` ${sampleExpr}";
@@ -177,6 +178,8 @@ public class HMSAnalysisTask extends BaseAnalysisTask {
         sb.append(ANALYZE_TABLE_TEMPLATE);
         Map<String, String> params = buildStatsParams("NULL");
         params.put("dataSizeFunction", getDataSizeFunction(col));
+        params.put("minFunction", getMinFunction());
+        params.put("maxFunction", getMaxFunction());
         StringSubstitutor stringSubstitutor = new StringSubstitutor(params);
         String sql = stringSubstitutor.replace(sb.toString());
         executeInsertSql(sql);

fe/fe-core/src/main/java/org/apache/doris/statistics/OlapAnalysisTask.java

@@ -81,11 +81,11 @@ public class OlapAnalysisTask extends BaseAnalysisTask {
             + "${idxId} AS idx_id, "
             + "'${colId}' AS col_id, "
             + "NULL AS part_id, "
-            + "COUNT(1) * ${scaleFactor} AS row_count, "
+            + "ROUND(COUNT(1) * ${scaleFactor}) AS row_count, "
             + NDV_SAMPLE_TEMPLATE
             + "SUM(CASE WHEN `${colName}` IS NULL THEN 1 ELSE 0 END) * ${scaleFactor} AS null_count, "
-            + "MIN(`${colName}`) AS min, "
-            + "MAX(`${colName}`) AS max, "
+            + "NULL AS min, "
+            + "NULL AS max, "
             + "${dataSizeFunction} * ${scaleFactor} AS data_size, "
             + "NOW() "
             + "FROM `${dbName}`.`${tblName}`"