Forecasting Method Analysis of Tin Ingots Company in Indonesia. Case Study: PT. Timah Tbk

Abstract

Mineral mining is performed globally, and mineral demand is continually growing in accordance with human necessities. Especially for Tin, the global tin market to grow at a CAGR of 0.81% during the period 2017-2021. PT. TIMAH Tbk, a state-owned company, operates the official tin mine on Bangka and Belitung Islands. The number of tin ingots is produced based on the RKAP or the planning master for PT. Timah Tbk. This research aims to get the well-performing forecasting method to make the estimation of tin ingots sales.

After analyzing the target and realization from 2009-2020, there several error between them with the specification MAD 7,291, MSD 123,983,550, and MAPE 18.1%. This error indicates that the target and realization have 18.1% different. In this research, there are several methods used to get the better forecasting method with lowest error. The methods conducted are based on (Nahmias & Olsen, 2015) time-series method analysis and causal method. The time-series analysis used are 4-Year Moving Average, Single Exponential Smoothing, Holt’s Model, and Winter’s Model. Furthermore, the causal method used are exercised by Support Vector Machine (SVM). The demand data used are based on sales of refined tins (tin ingots) from 2009-2020.

The result shows that SVM method or causal method has become the forecasting method with lowest forecast error with detail MAD 3,084, MSD 29,360,790, and MAPE 8.9%. This forecasting method is performing well because the factors used that affect the demand of tin in Indonesia have theoretically passed the correlation test (Safitri, 2004).

Keywords: Forecasting method analysis, tins company, PT. Timah Tbk.