Skip to main content

Table 3 Comparing the AUC results of Machine Learning algorithms with each balancing technique

From: Customer churn prediction in telecom using machine learning in big data platform

 Technique used for unbalanced dataset XGBOOST (%) GSM (B) (%) Random Forest (%) Decision Tree (%)
Oversampling 92 90.01 84.2 76.25
Undersampling 93.12 90.21 87.76 83
Without balancing 93.3 90.89 78.47 72.2
  1. As shown in the table each Machine Learning algorithm experimented with three different scenarios with regards to the problem of the unbalanced dataset. The best results of AUC presented in italics indicate the best technique that fits each algorithm