SUGGESTED PLAN TO IMPROVE SALES FORECAST ACCURACY BY USING TIME SERIES ANALYSIS AT PT BEKAERT INDONESIA – RUBBER REINFORCEMENT UNIT

• Main Author: Yurin Hazyanti, Dona
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/57610
• Institution: Institut Teknologi Bandung

PT. Bekaert Indonesia is a foreign investment company producing rubber reinforcement products. Based on an analysis of the business situation, research is focusing on the demand forecasting method. Forecasts are vital to every business organization and for every significant management decision. The existing forecasting error is above 5%, which consider the ideal forecast error and the average 3 years stock is above 2 times of sales beyond the management target. Higher inventory causing higher working capital spending. The current demand forecast method is also highly dependent on personnel. Jacob and Chase mention in their book Operation and Supply Chain Management, 15th edition, qualitative and time-series techniques are the most often use in supply chain planning and control. The test is conducted to whether one of the time series technique methods can be used to improve the forecast accuracy compared to the current qualitative method. The study compares all five forecasting methods of time series analysis. Below steps have been taken in this study. 1. Plotting sales volume by the country for 3 years, from 2018 to 2020 as most current historical sales forecast and actual. 2. Define the existing regular sales by country, choose only regular sales to study. 3. Understanding trends on historical sales volume versus inventory volume. 4. Choose the prominent sales by country to study 5. Cluster the customer in the country to have better focus and understanding of forecast behavior 6. Test the time series forecast on the chosen data 7. Do the result comparison 8. Calculate the possible saving Based on the study, the researcher can answer the two research questions below. 1. Did PT Bekaert Indonesia perform efficient demand forecasting? Based on the calculation, the mean absolute deviation and mean absolute percent error are still above a level that the researcher perceived as an ideal number, 100 ton and 5% in consecutive order. Conclude the current forecast method was not efficient enough for the company to have an optimum working capital. 2. What is the alternative forecasting method that is suitable for PT Bekaert Indonesia, which can be suggested to improves its working capital? Based on the test result, the researcher concludes that exponential smoothing with the trend can be suggested to replace the current forecasting method in PT Bekaert Indonesia. This suggested method still not reach an ideal targeted percentage error, need to conduct further test and study to have a better result on forecast deviation. The further study should put priority to systemize the quantitative forecast using time series analysis – exponential smoothing with trends for customers with high market share and low growth or has reached its maximum capacity, as this customer has stable conditions which demand can be predicted more accurate and the forecast accuracy has a high impact to the company. Follow-up activities need to be taken, emphasizing the urgent and important activities such as discussion and got alignment with the regional marketing manager on the follow-up studies.