Merge pull request #1031 from harsh-jindal/feature-abs-sum

Add abs sum as an aggregator
finos · May 8, 2020 · a4c28be · a4c28be
2 parents faa9f72 + 4c25d1f
commit a4c28be
Show file tree

Hide file tree

Showing 7 changed files with 20 additions and 1 deletion.
diff --git a/cpp/perspective/src/cpp/aggspec.cpp b/cpp/perspective/src/cpp/aggspec.cpp
@@ -98,6 +98,9 @@ t_aggspec::agg_str() const {
         case AGGTYPE_SUM_ABS: {
             return "sum_abs";
         } break;
+        case AGGTYPE_ABS_SUM: {
+            return "abs_sum";
+        } break;
         case AGGTYPE_MUL: {
             return "mul";
         } break;
@@ -281,6 +284,7 @@ t_aggspec::get_output_specs(const t_schema& schema) const {
     switch (agg()) {
         case AGGTYPE_SUM:
         case AGGTYPE_SUM_ABS:
+        case AGGTYPE_ABS_SUM:
         case AGGTYPE_PCT_SUM_PARENT:
         case AGGTYPE_PCT_SUM_GRAND_TOTAL:
         case AGGTYPE_MUL:

diff --git a/cpp/perspective/src/cpp/base.cpp b/cpp/perspective/src/cpp/base.cpp
@@ -590,8 +590,10 @@ str_to_aggtype(const std::string& str) {
         return t_aggtype::AGGTYPE_HIGH_WATER_MARK;
     } else if (str == "low" || str == "low_water_mark") {
         return t_aggtype::AGGTYPE_LOW_WATER_MARK;
-    } else if (str == "sum abs") {
+    } else if (str == "sum abs" || str == "sum_abs") {
         return t_aggtype::AGGTYPE_SUM_ABS;
+    } else if (str == "abs sum" || str == "abs_sum") {
+        return t_aggtype::AGGTYPE_ABS_SUM;
     } else if (str == "sum not null" || str == "sum_not_null") {
         return t_aggtype::AGGTYPE_SUM_NOT_NULL;
     } else if (str == "mean by count" || str == "mean_by_count") {

diff --git a/cpp/perspective/src/cpp/config.cpp b/cpp/perspective/src/cpp/config.cpp
@@ -193,6 +193,7 @@ t_config::setup(const std::vector<std::string>& detail_columns,
             case AGGTYPE_PY_AGG:
             case AGGTYPE_SUM_NOT_NULL:
             case AGGTYPE_SUM_ABS:
+            case AGGTYPE_ABS_SUM:
             case AGGTYPE_MUL:
             case AGGTYPE_DISTINCT_COUNT:
             case AGGTYPE_DISTINCT_LEAF:

diff --git a/cpp/perspective/src/cpp/extract_aggregate.cpp b/cpp/perspective/src/cpp/extract_aggregate.cpp
@@ -47,6 +47,7 @@ extract_aggregate(
         } break;
         case AGGTYPE_SUM:
         case AGGTYPE_SUM_ABS:
+        case AGGTYPE_ABS_SUM:
         case AGGTYPE_SUM_NOT_NULL:
         case AGGTYPE_MUL:
         case AGGTYPE_COUNT:

diff --git a/cpp/perspective/src/cpp/sparse_tree.cpp b/cpp/perspective/src/cpp/sparse_tree.cpp
@@ -1192,6 +1192,15 @@ t_stree::update_agg_table(t_uindex nidx, t_agg_update_info& info, t_uindex src_r
                         }));
                 dst->set_scalar(dst_ridx, new_value);
             } break;
+            case AGGTYPE_ABS_SUM: {
+                old_value.set(dst->get_scalar(dst_ridx));
+                auto pkeys = get_pkeys(nidx);
+                std::vector<double> values;
+                gstate.read_column(spec.get_dependencies()[0].name(), pkeys, values);
+                double sum = std::accumulate(values.begin(), values.end(), double(0));
+                new_value.set(std::abs(sum));
+                dst->set_scalar(dst_ridx, new_value);
+            } break;
             case AGGTYPE_MUL: {
                 old_value.set(dst->get_scalar(dst_ridx));
                 auto pkeys = get_pkeys(nidx);

diff --git a/cpp/perspective/src/include/perspective/base.h b/cpp/perspective/src/include/perspective/base.h
@@ -305,6 +305,7 @@ enum t_aggtype {
     AGGTYPE_UDF_COMBINER,
     AGGTYPE_UDF_REDUCER,
     AGGTYPE_SUM_ABS,
+    AGGTYPE_ABS_SUM,
     AGGTYPE_SUM_NOT_NULL,
     AGGTYPE_MEAN_BY_COUNT,
     AGGTYPE_IDENTITY,

diff --git a/python/perspective/perspective/core/aggregate.py b/python/perspective/perspective/core/aggregate.py
@@ -35,6 +35,7 @@ class Aggregate(Enum):
     PCT_SUM_GRAND_TOTAL = 'pct sum grand total'
     SUM = 'sum'
     SUM_ABS = 'sum abs'
+    ABS_SUM = 'abs sum'
     SUM_NOT_NULL = 'sum not null'
     UNIQUE = 'unique'