Blood Bag Stock Availability Forecasting System At PMI UDD Lhokseumawe City Using The Random Forest Method

Putri, Rizka Hilmi

Electronic Journal of Education, Social Economics and Technology
Vol 6, No 2 (2025)

This study aims to develop a blood bag stock forecasting system at the Indonesian Red Cross (PMI) Blood Donor Unit (UDD) of Lhokseumawe City and PMI North Aceh UDD using the Random Forest (RF) algorithm. The background of this research focuses on the imbalance between the demand and availability of blood bags, which can have serious implications for patient safety. Using historical data of 574 records from the period 2022 to 2025, this study adopted a quantitative approach involving variables such as month, year, PMI location, blood type, incoming blood, and blood distribution. The research method included preprocessing and feature engineering steps, which resulted in 17 features for Random Forest (RF) model training. This process was completed with hyperparameter optimization using RandomizedSearchCV to improve model accuracy. The results showed that UDD PMI North Aceh achieved excellent performance with an RMSE of 9.113 for incoming blood and 8.750 for distribution. In contrast, UDD PMI Lhokseumawe had RMSEs of 24,635 and 22,737, respectively. Comprehensive predictions for the period 2025-2030 showed optimistic projections with 94% surplus months and a positive net balance of 4,406.7 bags. The implemented web-based system supports real-time forecasting and strategic decision-making in blood bag stock management at PMI. Keywords: Forecasting; Blood; Transfution; Stock; RF