Forecasting latex crepe prices in the Sri Lankan market: A comparison of conventional time series models with artificial neural networks (ANN) (Record no. 75224)

MARC details
000 -LEADER
fixed length control field 02098nam a22001577a 4500
100 ## - MAIN ENTRY--AUTHOR NAME
Personal name Wijesuriya, W.
245 ## - TITLE STATEMENT
Title Forecasting latex crepe prices in the Sri Lankan market: A comparison of conventional time series models with artificial neural networks (ANN)
260 ## - PUBLICATION, DISTRIBUTION, ETC. (IMPRINT)
Name of publisher Proceedings of the IRRDB International Rubber Conference 2023, 20-21 February 2023, IRRDB, Kuala Lumpur, Malaysia, pp. 107-112.
520 ## - SUMMARY, ETC.
Summary, etc This study attempted to forecast Latex Crepe Grade 1 (LC-1) prices of the Sri Lankan market through Seasonal Auto Regressive Integrated Moving Average (SARIMA), Holt Winter’s exponential smoothing (HW_ETS) and Artificial Neural Network (ANN) models. A monthly time series data of LC-1 from January 1986 to December 2020 were used for estimation of LC-1 prices with SARIMA and HW_ETS. For Artificial Neural Network (ANN) models, 70% of the data set was used for training and the rest was used for testing. To forecast LC-1 prices, three time periods viz. three, six and 12 months were used for all models beyond January 2021. The variables, international rubber prices (IntP), exchange rates, Sri Lankan rupee per US$ (SLR_$) and Japanese Yen per US$ (Yen_$) and crude oil prices (Cr_OP) were used with their first two lags as independent variables in ANN, together with 12 lags of monthly LC-1. The forecast performance of the models was evaluated using Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). The selected ANN models (ANN_M1: 12 lags of LC-1 and ANN_M2: 12 lags of LC-1 with Cr_OP and its first 2 lags) yielded relatively better forecast performances than the conventional time series models and the other ANN models tested. For 12 and six months forecasts, model M_1 performed better and for three months forecasts, the model M_2 was found to be the best model with acceptable ranges of MAPE (8-9%) and RMSE (62-63).
650 ## - SUBJECT ADDED ENTRY--TOPICAL TERM
Topical Term Artificial Neutral Networks (ANNs)
650 ## - SUBJECT ADDED ENTRY--TOPICAL TERM
Topical Term Latex crepe prices
650 ## - SUBJECT ADDED ENTRY--TOPICAL TERM
Topical Term Price forecasting
650 ## - SUBJECT ADDED ENTRY--TOPICAL TERM
Topical Term Time series models
700 ## - ADDED ENTRY--PERSONAL NAME
Personal name Rathnayaka, D., Sankalpa, S. and Ishani, P.G.N.
942 ## - ADDED ENTRY ELEMENTS (KOHA)
Koha item type Journals
Holdings
Withdrawn status Lost status Damaged status Not for loan Home library Current library Date acquired Koha item type
        RRII Library RRII Library 25/04/2024 Journals