C- ok Yan-Tl- Doǧrusal Regresyon Algoritmas-yla Lojistik Regresyon ve Birliktelik Kural- Sonuçlarinin Birleştirilmesi


Ekmekçi N., Kasap Ö. Y. , Ketenci U. G. , Kalipsz O., AKTAŞ M. S.

10th Turkish National Software Engineering Symposium, UYMS 2016, Çanakkale, Turkey, 24 - 26 October 2016, vol.1721, pp.611-622 identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 1721
  • City: Çanakkale
  • Country: Turkey
  • Page Numbers: pp.611-622
  • Keywords: Association rules, Data mining, Ensemble learning, Logistic regression, Multi-response linear regression, Stacking

Abstract

One of the keys in marketing is to recommend the right products to the right customers. This paper proposes a solution to this problem as a part of the development of a new data mining tool PROPCA. The aim is to use logistic regression analysis and association rule mining together to make recommendations in marketing. An approach in which combination of these two algorithms provides better results than algorithms used standalone is presented. While association rule mining searches all rules in the data set, logistic regression predicts a purchase probability of a product for customers. The combination of these two approaches are tested on a real-life banking data set. The results of combination are shown and their suitability in general is discussed.