Demand Forecasting and Capacity Planning of Polyester Filament Yarn in Taiwan

Abstract

Taiwan's man-made fiber industry plays an extremely important role in the development of Taiwan’s textile industry providing raw materials to spinning industry weaving industry and apparel industry Polyester Filament Yarn the most important product in Taiwan's man-made fiber industry its annual yield accounts for up to 50 percent of all man-made fibers in Taiwan However the demand uncertainty of Polyester Filament Yarn caused by the drastic change of international situations including the rise of Southeast Asian countries economy the fluctuations of crude oil price the rise of environmental awareness and U S -China trade war etc If the demand for Polyester Filament Yarn can be predicted accurately Taiwan’s textile companies can make a better capacity planning and prepare for the future changes The study is divided into two stages the first stage is to predict the demand of Polyester Filament Yarn in Taiwan by using Autoregressive Integrated Moving Average model (ARIMA model) and Univariate/ Multivariate Long Short-Term Memory models (LSTM model) adding relevant variables such as raw material price and competitive product price to establish forecast models; the second stage continues the first stage forecast result for capacity planning adding several cost variables to establish Linear Programming model to minimize costs hoping to provide a solution for Taiwan’s textile companies under demand uncertainties The study shows that the forecasting result by Multivariate LSTM model is more accurate than other models As for capacity planning the profit will increase 1 2% by cost minimization

Date of Award	2020
Original language	English
Supervisor	Chia-Yen Lee (Supervisor)