Indonesia Conference Directory

This study aims to model the form of customer profile classification on companies using the C 4.5 and Random Forest algorithms to produce the best profile classification model from customers to sess a pattern of assessments of manual assessments so far. The general model used in customer grouping is the RFM (Recency, Frequency, Monetary) model, which is grouped based on the latest customer visits, the frequency of visits, and income received by the company, but in this case there is C (Cost) as new attribute, so the customer grouping model becomes RFM-C model approach. The use of CRISP-DM data mining methodology is the methodology used in this study, where the design of the C4.5 algorithm and Random Forest algorithm is compared and validated using 10-fold validation, in this case the application used is RapidMiner. While the 172,261 customer transaction data collected for two years. After testing the two algorithms, the results obtained are the use of the C4.5 algorithm for companies to classify RFM-C which is expected to predict because it has higher accuracy and kappa. And now these plastic packaging companies can predict customer transaction data based on historical data that has been processed.