We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
1 parent 33d5894 commit 4f36f27Copy full SHA for 4f36f27
README.md
@@ -20,6 +20,13 @@
20
- [x] Version 3: CMTR_CHURN_PR_V3_ TabNet 使用无监督模型进行预训练
21
- 遇到的问题:AUC和准确率提升依旧很难,模型效果比较差。AUC: 0.6511318268787747 | Score: 0.74
22
- 解决思路:使用PCA主成分分析法,进行特征降维, 准确率依旧下降,出现特征工程无效的情况,原因未知。AUC: 0.5801404503216607
23
+-----
24
+- [x] Version 3: CMTR_CHURN_PR_V4 变量分箱 | XGB | RF | TabNet | AutoML
25
+ - 经过特征选择后(TOP30),对字符型变量进行独热编码,对数值型去除量纲,并进行等间距分箱,num_bins = 6
26
+ 1. XGB: AUC: 0.6046600198503995 | Score: 0.775887943971986
27
+ 2. RandomForest: AUC: 0.6046600198503995 | Score: 0.7712606303151576
28
+ 3. TabNet 无监督预训练, AUC: 0.7872822945848122
29
+ 4. AutoGluon 直接输出leaderboard,并使用TOP3的模型进行预测。
30
31
32
0 commit comments