|
|
@@ -7,7 +7,9 @@
|
|
|
# Distributed under terms of the MIT license.
|
|
|
|
|
|
"""
|
|
|
-删除容易导致偏差的viewall特征
|
|
|
+1.删除容易导致偏差的viewall特征
|
|
|
+2.删除分桶不均匀的cpa特征
|
|
|
+3.减少dense特征
|
|
|
"""
|
|
|
|
|
|
|
|
|
@@ -25,10 +27,10 @@ def read_features(filename, excludes=None):
|
|
|
features.remove(x)
|
|
|
return features
|
|
|
|
|
|
-exclude_features = ['viewall',]
|
|
|
+exclude_features = ['viewall', 'cpa']
|
|
|
|
|
|
-dense_features = read_features("features_top300.config", exclude_features)
|
|
|
-top_dense_features = read_features('features_top100.config', exclude_features)
|
|
|
+dense_features = read_features("features_top300.config", exclude_features)[:150]
|
|
|
+top_dense_features = read_features('features_top50.config', exclude_features)
|
|
|
|
|
|
sparse_features = [
|
|
|
"cid", "adid", "adverid",
|
xueyiming