BROWSE

ITEM VIEW & DOWNLOAD

A survey of the optimal time series modelling for newbuilding ship prices

Cited 0 times inthomson ciCited 0 times inthomson ci
Title
A survey of the optimal time series modelling for newbuilding ship prices
Author
Jeon, Jaemin
Advisor
Woo, Hangyun
Keywords
LSTM; Savitzky–Golay filter; Newbuilding Price Index; VECM
Issue Date
2020-02
Publisher
Graduate School of Technology and Innovation Management
Abstract
The shipping industry can be divided into four markets : the freight market, the newbuilding market, the second-hand, and the demolition market. The newbuilding market is especially responsible for connecting two pillars of maritime business—the shipping and shipbuilding industries—with newbuilding ship prices. The newbuilding ship prices are the sum of measured values of vessels to be constructed at the time of contracting with shipowners, and newbuilding ship price market forecasting would be a criterion of strategic decision-making in shipyards. Therefore, a reasonable estimation of the newbuilding ship price can be a driver for growth in shipyard management. Previous studies on the determining factors for newbuilding ship prices are rare, and some of the work is old and requires reinvestigated. Also, since the newbuilding market is volatile, time-series forecasting methodologies that assume linearity have limitations in terms of utilization. To propose an optimal newbuilding ship price estimation model, we built and compared Vector Error Correction Model (VECM) Long Short-Term Memory (LSTM) with hyper-parameter optimization. Through a literature review, we selected economic variables, including second-hand ship prices, freight rates, and interest rates from January 1986 to June 2019, and verified their influence on newbuilding ship prices. For the validation and evaluation of the time-series models, we conducted a sliding window test to achieve prediction robustness. As a result, we empirically confirmed the superiority of LSTM based on neural network that revealed better performance in rapidly changing periods. Additionally, we applied a Savitzky–Golay filter that eliminates noises from time-series variables and combined it with the forecasting models, and the experimental results indicate that models that are integrated with denoising filters exhibit better performance than single models. Based on empirical tests in this study, we propose a time-series forecasting model combining the Savitzky–Golay filter and LSTM in the newbuilding ship price market.
Description
Department of Technology and Innovation Management
URI
Go to Link
Appears in Collections:
SBA_Theses_Master
Files in This Item:
A survey of the optimal time series modelling for newbuilding ship prices.pdf Download

find_unist can give you direct access to the published full text of this article. (UNISTARs only)

Show full item record

qrcode

  • mendeley

    citeulike

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

MENU