With the rapid development of mobile communication and Internet technology, operators' data services are bringing scientific breakthroughs that can produce chain reaction, resulting in a number of innovative Internet products. With the increasing demands of operator's products and services, the amount of user-related complaints is also growing rapidly. This not only puts negative impacts on operator business innovation, but also leads to a continuous decrease in user satisfaction. In order to address this issue, we propose a user satisfaction assessment based on signaling mining. This assessment analyzes the signaling data of the users who complain at different time and performs regionalization. Finally, a user complaint prediction model based on GBDT(Gradient Boosting Decision Tree) algorithm is established to output the user complaint probability. The experimental results show that our proposed model is 15% higher than the state-of-art complaint models on the AUC indicator. This provides a positive data reference for predicting user complaints, identifying high-risk users, and improving user satisfaction.
[1]
Ajinkya More,et al.
Survey of resampling techniques for improving classification performance in unbalanced datasets
,
2016,
ArXiv.
[2]
M. Zweig,et al.
Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine.
,
1993,
Clinical chemistry.
[3]
Leo Breiman,et al.
Random Forests
,
2001,
Machine Learning.
[4]
J. Friedman.
Greedy function approximation: A gradient boosting machine.
,
2001
.
[5]
Chih-Jen Lin,et al.
LIBLINEAR: A Library for Large Linear Classification
,
2008,
J. Mach. Learn. Res..
[6]
Peter Schmidt,et al.
Key performance indicators versus business process metrics
,
2018,
Int. J. Adv. Oper. Manag..
[7]
Isabelle Guyon,et al.
Design and analysis of the KDD cup 2009: fast scoring on a large orange customer database
,
2009,
SKDD.