Data Mining Approach for Analyzing Call Center Performance

Title: Data Mining Approach for Analyzing Call Center Performance

Abstract: This research aimed to apply data mining techniques to predict call center quality of service based on performance data. The study used various methods like linear neural networks, multi-layered perceptrons, probabilistic neural networks, classification and regression trees, support vector machines, and a hybrid decision tree-neural network approach. The findings showed that the hybrid approach performed best, and sensitivity analysis revealed that call duration and customer satisfaction were the most significant factors affecting call center performance. These insights can help improve call center performance.

Main Research Question: Can data mining techniques be used to predict call center quality of service based on performance data?

Methodology: The study used data from a large insurance company's call centers. It applied different data mining techniques to the collected data and compared their performance. The researchers also conducted sensitivity analysis to understand the input variables' impact on the prediction models.

Results: The hybrid decision tree-neural network approach outperformed the other techniques. Sensitivity analysis revealed that call duration and customer satisfaction were the most significant factors affecting call center performance.

Implications: The study's findings suggest that data mining techniques can be used to predict call center quality of service. The hybrid decision tree-neural network approach was found to be the most effective. Additionally, the sensitivity analysis highlighted the importance of call duration and customer satisfaction in determining call center performance. These insights can help improve call center performance by focusing on these key factors.

Link to Article: https://arxiv.org/abs/0405017v1 Authors: arXiv ID: 0405017v1