DECISION SUPPORT SYSTEM FOR THE NUMBER OF BLOOD REQUESTS PREDICTION AT THE BLOOD DONOR UNIT PMI JEMBER USING LINEAR REGRESSION AND DOUBLE EXPONETIAL SMOOTHING METHODS
One in four people in the world needs a blood transfusion during their lifetime, but only 37% of the population qualify as donors, and only 10% donate blood regularly. The fact that more blood needs than donating blood makes UDD PMI challenging to meet demand when the existing bloodstock is insufficient or empty. And based on the results of research interviews with sources, there was a problem of increasing blood stocks that decreased in certain months, followed by a decrease in blood demand, which caused a reasonably high change in the following month. This caused blood demand could not be met due to reduced demand and supply but increased in the next month without sufficient stock increase. Based on the problems described above, researchers see the need to estimate the number of blood demand needs in the Jember PMI UDD with linear regression and Double Exponential Smoothing methods. With the forecasting and understanding of past time series models, it is possible to predict future values. From there, researchers want to minimize stakeholder decision-making errors and provide options to stakeholders concerning the minimum number of blood bags that need to be provided. In conclusion, the results of the prediction implementation predict data that is different from the data found in the field. The approximate accuracy is 60.32% and 46.96% for linear regression and double exponential smoothing. The worst absolute error value of the linear regression method is data 0 and 15 using actual data. The worst absolute error is the worst double exponential data 0 and 27.7.
Keywords: Blood transfusion, demand and supply, forecasting, linear regression, Double Exponential Smoothing