Feature Selection for Campaign Success Prediction

Businesses that are going digital accumulate large volumes of customer data every day, and one of the most valuable uses of that data is customer analytics. On the CRM side, being able to foresee whether a campaign will succeed before it even starts has a direct effect on budget and target audience decisions. This paper describes the pilot study the Next4biz R&D team carried out on exactly that question.

In the study, descriptive data from campaigns previously run on the Next4biz CRM software was brought together in a single data set. Before building a prediction model, the team conducted a feature selection study to determine which features genuinely carry information, and compared the results. The goal was to improve the development stage of the prediction model through feature selection. The paper's reference list cites Boruta, random forest, and correlation- and information-gain-based selection methods; the published abstract, however, does not state which method stood out or report a numerical performance figure.

The paper was presented by Emrah Sezer, Ahmet Erkan Çelik and Teoman Berkay Ayaz at the 9th International Conference on Future Learning and Informatics (October 2022) and appeared in the congress abstract book. Its keywords are Feature Selection, Machine Learning and Customer Relationship Management. The team's work on campaign success prediction continued in the following years with topics such as the effect of hyperparameter tuning.

Full text of the paper: Open the PDF in a new tab

Prof. Dr. Akhan Akbulut

Professor Doctor Akhan Akbulut worked in the Computer Engineering departments of Istanbul Kültür University and NC State University. He serves institutions such as TÜBİTAK, Ministry of Industry and Technology, TÜSEB, and KOSGEB. He researches Distributed Systems and Artificial Intelligence and has over 100 international journal articles and conference proceedings from his work.