İLAÇ SEKTÖRÜNDE ZAMAN SERİSİ VE REGRASYON BİRLEŞİK MODELLER İLE TALEP TAHMİNİ UYGULAMASI

Abstract

Accurate demand forecasting is crucially important to reduce inventory and backlogging cost. In this study, we analyze how promos, holiday statements, price changes, stock availability and date-time features (weekdays, months etc.) affect the demand by using several forecasting methods. Data sets were collected for the products of the global pharmaceutical company providing services in Turkey. Actual daily sales data for 2016, 2017 and 2018 were used in the construction of this data set. In order to predict the next periods demand, we used four different models such as Holt Winters, Ridge Regression, Random Forest and Xgboost. We also ensemble those models to improve forecasting accuracy. Our numerical results show that the lowest forecasting error rate was obtained in ensemble models. Particularly, the lowest error rate in individual models was obtained in Random Forest with 15,7% RMSPE (Root Mean Percentage Value) value, and the lowest error rate was obtained with 10.7% RMSPE value in Holt Winters & Xgboost combination models. Moreover, the correlation coefficients of the features between sales are also presented.