Hi all, here is the abstract for metal price prediction. My supervisor advised me to rewrite for bad grammar. Hope you guys can make comments for that. THANKS!
Metal commodity price is crucial in the profitability equation for mining industry. If assuming metal cost remain the same, profits will be expected to ascend with increasing producer prices. Thus, forecasting tools are important for producers to increase stability and predictability of metal commodity price. This thesis analyses the ability of two time series predicting models to forecast price for aluminium, copper, gold, zinc, nickel and lead based on observed historical data. Additionally, this thesis compares model outputs with predicted results from market to observe accuracy of two time series models.
This thesis applies Auto Regressive Moving Average (ARMA) and Autoregressive Integrated Moving Average (ARIMA) time series model to predict the future metal commodity price based on past data from January 1990 to March 2018. The ARMA and ARIMA models with the lowest Akaike's Information Criterion (AIC) are selected from each series. The results show that the predicted value of model is highly fitting to the historic price in 4 months. However, ARMA and ARIMA are based on assumption of linear historic metal price and used for short run only, to detect small variation between the data.
In real world, financial institutions such as 'World Bank' (my supervisor put comment 'plus many others' for 'World Bank') provides historic metal price and predicted price curve to public. Mining company will consider metal commodity price from market for maximizing mining profit. Thus, the accuracy of metal commodity price predictor in public is initially important for mining industry. The limitation of this thesis is simply predicting and analysing the metal commodity price by using time series models without any economic and politic factors from the market.
Key words: Forecast price, time series, ARMA, ARIMA, accuracy
Forecast price, time series, ARMA/arima, accuracy
Metal commodity price is crucial in the profitability equation for mining industry. If assuming metal cost remain the same, profits will be expected to ascend with increasing producer prices. Thus, forecasting tools are important for producers to increase stability and predictability of metal commodity price. This thesis analyses the ability of two time series predicting models to forecast price for aluminium, copper, gold, zinc, nickel and lead based on observed historical data. Additionally, this thesis compares model outputs with predicted results from market to observe accuracy of two time series models.
This thesis applies Auto Regressive Moving Average (ARMA) and Autoregressive Integrated Moving Average (ARIMA) time series model to predict the future metal commodity price based on past data from January 1990 to March 2018. The ARMA and ARIMA models with the lowest Akaike's Information Criterion (AIC) are selected from each series. The results show that the predicted value of model is highly fitting to the historic price in 4 months. However, ARMA and ARIMA are based on assumption of linear historic metal price and used for short run only, to detect small variation between the data.
In real world, financial institutions such as 'World Bank' (my supervisor put comment 'plus many others' for 'World Bank') provides historic metal price and predicted price curve to public. Mining company will consider metal commodity price from market for maximizing mining profit. Thus, the accuracy of metal commodity price predictor in public is initially important for mining industry. The limitation of this thesis is simply predicting and analysing the metal commodity price by using time series models without any economic and politic factors from the market.
Key words: Forecast price, time series, ARMA, ARIMA, accuracy