Data Processing for Direct Marketing Through Big Data Amelec Viloria 1(&) , Noel Varela 1 , Doyreg Maldonado Pérez 1 , and Omar Bonerge Pineda Lezama 2 1 Universidad de la Costa (CUC), Calle 58 # 55-66, Baranquilla, Atlantico, Colombia {aviloria7,nvarela1,dmaldona}@cuc.edu.co 2 Universidad Tecnológica Centroamericana (UNITEC), San Pedro Sula, Honduras omarpineda@unitec.edu Abstract. Traditional marketing performs promotion through various channels such as news in newspapers, radio, etc., but those promotions are aimed at all people, whether or not interested in the product or service being promoted. This method usually leads to high expenses and a low response rate by potential customers. That is why, nowadays, because there is a very competitive market, mass marketing is not safe, hence specialists are focusing efforts on direct marketing. This method studies the characteristics, needs and also selects a group of customers as a target for the promotion. Direct marketing uses pre- dictive modeling from customer data, with the aim of selecting the most likely to respond to promotions. This research proposes a platform for the processing of data flows for target customer selection processes and the construction of required predictive response models. Keywords: Data stream Á WEKA Á MOA Á SAMOA Á Big Data Á Direct marketing 1 Introduction Direct marketing is a data set-oriented process for direct communication with target customers or prospects [1, 2]. But at the same time, marketers face the situation of changing environments. Current datasets constitute computers insert data into each other making environments dynamic and conditioned by restrictions such as limited storage capacity, need for real-time processing, etc. This means that Big Data processing approaches [3, 4] are required; Hence, in order to establish and maintain the relationship with clients, specialists have anticipated the need to change the methods of intuitive group selection for more scientific approaches aimed at processing large volumes of data [5], obtaining rapid responses that allow selecting customers who will respond to a new offer of products or services, under a distributed data flow approach [6]. Several researches relate to computational and theoretical aspects of direct mar- keting, but few efforts have focused on technological aspects needed to apply data mining to the direct marketing process [7–11]. Situation is that gains in complexity © Springer Nature Switzerland AG 2020 S. Smys et al. (Eds.): ICCVBIC 2019, AISC 1108, pp. 187–192, 2020. https://doi.org/10.1007/978-3-030-37218-7_21