Optimizing the prediction of telemarketing target calls by a classification technique

Walid Cherif, Hassan Silkan

2018 · 20 citations · 13 references

Concepts

Abstract

This paper presents a new classification technique to optimize the prediction of telemarketing target calls for selling bank long-term deposits. A Portuguese retail bank addressed, from 2008 until 2013, data on its clients, products and social-economic attributes including the effects of the financial crisis. An original set of 150 features has been explored and 21 features are retained for the proposed approach. This paper introduces a new technique that implicitly fosters most significant features and predicts the classes of clients according to the types of these features. Moreover, experiments showed that either these features are normalized or not, the proposed technique proved stable and accurate. To evaluate the obtained results, this paper compares them to those of most known machine learning models: Naive Bayes (NB), Decision Trees (DT), Artificial Neural Network (ANN) and Support Vector Machines (SVM). Thus, the proposed approach yielded the best performance in terms of f-measure, and it allowed reaching 60.12% of the subscribers.

References

13