Changeset 1519 for trunk/yat/normalizer/QuantileNormalizer.cc

Timestamp:

Sep 21, 2008, 6:35:39 AM (13 years ago)

Author:

Peter

Message:

speeding up QuantileNormalizer? by avoiding some copying and by only sorting each column once and not twice as before.

File:

• trunk/yat/normalizer/QuantileNormalizer.cc (modified) (1 diff)

trunk/yat/normalizer/QuantileNormalizer.cc

@@ -41 +41 @@
     assert(data.columns()==result.columns());
 
-    // create a tmp copy
-    utility::Matrix data_copy(data);
-    
-    // sort columns in copy
-    for (size_t column=0; column<data_copy.columns(); ++column){
-      std::sort(data_copy.begin_column(column), data_copy.end_column(column));
-    }
-
-    // calculate average of each row
-    std::vector<yat::statistics::Averager> averager(data_copy.rows());
-    for (size_t row=0; row<result.rows(); ++row){
-      add(averager[row], data_copy.begin_row(row), data_copy.end_row(row));
-    }
-
-    for (size_t column=0; column<result.columns(); ++column){
-      std::vector<size_t> index;
-      utility::sort_index(index, data.column_const_view(column));
-                               
-      for (size_t row=0; row<data.rows(); ++row)
-        result(index[row], column) = averager[row].mean();
+    std::vector<std::vector<size_t> > index(data.rows());
+    for (size_t column=0; column<data.columns(); ++column)
+      utility::sort_index(index[column], data.column_const_view(column));
+      
+    for (size_t rank=0; rank<data.rows(); ++rank) {
+      statistics::Averager a;
+      for (size_t column=0; column<data.columns(); ++column)
+        a.add(data(index[column][rank], column));
+      double mean = a.mean();
+      for (size_t column=0; column<data.columns(); ++column)
+        result(index[column][rank], column) = mean;
     }
   }