Classification Algo to Predict the Success of Telemarketing Campaign

Abstract

introduction: Marketing plays a vital role when it comes to the success of any business. Companies use direct marketing as a tool to target the customers and grab their attention. In order to ease the operational management campaign, the customers’ remote interaction is centralized in a contact center. With the help of such centers customers are contacted via different channels, telephone being one of the most widely used. The marketing which is operationalized through contact center is referred as telemarketing. Contacts are divided into two broader categories inbound and outbound, depending on who initiated the contact. Technology enables the marketers to evaluate the profile of customers and on the basis of it identity the potential clients that are more likely to purchase the product or services in future. DSS (Decision Support System) makes use of information technology in order to support managerial decision making. DSS is further divided in to sub fields: Personal DSS and intelligent DSS. Personal DSS is used on a small scale where as Intelligent has the ability to support various decisions simultaneously. There is another related concept Business intelligence which includes information technologies such as Data mining and data warehouses used to support decision making. Data mining plays a vital role in personal and intelligent DSS. The most common DM task is classification and the underlying goal is to make model that is driven by data set in order to learn unknown underlying function to map different input variables in order to characterize an item with the target output. We have used the Data mining approach to find out the success of telemarketing campaign. Several classification models are available such as Logistic Regression (LR) Decision Tree (DT), Neutral Network (NW) and Support Vector Machines (SVM). LR and DT are easily understandable ad provide good prediction regarding classification tasks whereas NW and SVM are more advanced they range from linear to complex nonlinear mapping.