Prediksi Churn Donatur Lembaga Swadaya Masyarakat Menggunakan Metode Random Forest Classifier

Abstract

Rapid technological developments have made machine learning (ML) an important tool in business and organizational operations. ML allows computer systems to perform tasks without explicit instructions, relying on patterns and inference from historical data to produce accurate predictions. The commercial sector has widely adopted ML to improve performance. Non-Governmental Organizations (NGOs) are community-based organizations that focus on improving and developing society. NGOs rely heavily on external donations for their financial sustainability, but often face challenges in retaining donors. A humanitarian NGO in North Sumatra is facing challenges in donor retention, with 93% of donors categorized as inactive (churn), indicating challenges in obtaining regular donations. This instability threatens the survival of the organization. The application of ML in NGOs allows institutions to predict donor behavior by analyzing historical data, thereby enabling NGOs to take action on the basis of predictions. This research produces a model based on the Random Forest Classifier classification method. The results of the classification analysis are that the Random Forest Classifier method shows an accuracy level of 88.3%, precision 92.0%, recall 85.5% and an AUC value of 96.6% so that it can be categorized as an excellent classification model