PERANCANGAN MODEL PREDIKSI KEPUTUSAN PURCHASE PENDAFTAR PROGRAM FREE TRIAL DI PT X DENGAN MENGGUNAKAN METODE DATA MINING

• Main Author: Naufal F Sahab, M
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55985
• Institution: Institut Teknologi Bandung

A free trial program is one of the highest revenue contributor in PT X, an educational company that offers study abroad preparation services. However, the number of customers generated from this program is still far from what is expected. The closing rate of free trial participants is still far below the ideal level, even tends to be low and decreasing compared to other programs such as webinars. In fact, the free trial program has succeeded in attracting many interested people, forcing the registrant to go through a selection process based on their personal information. The opportunity loss and the problem were happened due to the absence of a decision-making system for the selection of applicants. To overcome this problem, in this study, a data-mining model was designed to predict the purchase decision of free trial registrants so that PT X can increase its closing rate and minimize decision-making time. An application prototype later designed so that the model can be used by the marketing division for the selection process. The design of the solution in this study refers to the CRISP-DM methodology. There are three classification algorithms used to build the prediction model, namely Support Vector Machine (SVM), Random Forest, and Gradient Boosting. Then, to handle the imbalanced data, two oversampling techniques were used, namely ADASYN and SMOTE. The best model obtained in this study is the Gradient Boosting model that was applied to the SMOTE data, with an accuracy value of 0.83, precision 0.5, recall 0.33, and F1 0.4. Applying the model, PT X could expect their closing rate to raise until 50%.