PREDICTION OF LATE PAYMENT CUSTOMER'S ELECTRICITY ACCOUNT USING MACHINE LEARNING

• Main Author: Puspita Sari Nilam Utami, Dyah
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86929
• Institution: Institut Teknologi Bandung

Electricity is a vital necessity in modern life, and managing electricity bill payments is crucial for the sustainability of services and the financial stability of electricity providers like PLN. Identifying potential late payments by customers is a strategic step to enable effective preventive measures. This study develops a predictive model for late payments using two machine learning methods, namely Random Forest and Bidirectional Long Short-Term Memory (LSTM), based on historical customer data from 2018–2023. The research process involved data preprocessing to ensure consistency and accuracy, splitting the data into training and testing sets, and training the models using both algorithms. The Random Forest model demonstrated the best performance in identifying long-term statistical patterns with the lowest Mean Absolute Error (MAE) of 0.00387 and 99,96% accuracy when using a 12-month Moving Average feature. This algorithm also showed optimal efficiency with a tree count between 100–200, delivering stable predictions without compromising computational time. On the other hand, the Bidirectional LSTM model exhibited competitive performance in capturing the temporal patterns of sequential data. The best configuration (learning rate 0.00005, batch size 8, epoch 100) achieved a validation error of 0.243 and the highest validation accuracy of 56.2%. This model excels in learning complex patterns, although it requires precise hyperparameter tuning to reduce the risk of overfitting. The study’s results show that both models are effective in predicting customers at risk of late payments. This research provides significant contributions to PLN by supporting data-driven decision-making, enabling strategies such as early notification or rescheduling payment plans to mitigate the risk of payment arrears.