PREDIKSI KEBUTUHAN OBAT DENGAN SUPPORT VECTOR REGRESSION DAN LONG SHORT TERM MEMORY

• Main Author: Novriza Alam, Ekky
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43460
• Institution: Institut Teknologi Bandung

Pharmacy Unit is a unit of provider and management of medicine in a hospital, clinic or puskesmas unit. To ensure the patient's needs related to medicine can be met, the pharmaceutical unit must manage drug storage properly by implementing inventory management. Health service unit inventory management must take precautions so that drug stocks are always at safe amounts. During this time, the majority of health service units such as hospitals, health centers, and clinics in Indonesia still use the traditional way of using instinct to predict the amount of needs so that stock is always available if needed. This research develops a method for predicting drug needs with the learning algorithm Support Vector Regression (SVR) and Long Short-Term Memory (LSTM). The data used are transaction data of drug needs during 2018. Prediction methods will produce a model of each learning. The model will be compared using RMSE so that the best model for each drug is obtained. The results of this study indicate that LSTM on weekly prediction has a smaller RMSE value compared to daily prediction (Demand Smooth and Erratic). LSTM learning compared to SVR learning has a smaller RMSE value except: Daily prediction: erratic demand and weekly prediction: lumpy demand.