Forecasting Ethiopian Agricultural Commodity Price Using Time Series Features and Technical Indicators
No Thumbnail Available
Date
2022-05
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Agricultural commodity price prediction helps the government, investors, and farmers
to make informed decisions. Realizing the benefit, several researchers proposed different
prediction models that use different features. However, most prediction models are
affected by factors, such as data type (e.g., linear and nonlinear), seasonality of commodity
items, weather conditions, commodity volatility features, and country economic
factors. Among these factors, the most significant impediments to the accuracy of commodity
price prediction are seasonality and trend pattern. To fill this gap, we propose
a model that predicts commodity prices through the combination of time series features
and technical indicators. The prediction model is built using four-machine learning algorithms:
Artificial Neural Network, Extreme Learning Machine, Support Vector Machine,
and Random Forest. To assess the impact of the proposed approach, we conducted two
experiments using coffee and sesame datasets. The performance of the prediction models
is assessed using the root mean square error (RMSE) and mean average error (MAE). The
results show that the proposed approach improves agricultural commodity price prediction
performance in all cases except MAE of sesame while using Extreme Learning
Machine. Using Artificial Neural Network, Extreme Learning Machine, Support Vector
Machine and Random Forest, the RMSE of price prediction is reduced by an average of
4.37, 4.42, 2.74, and 5.15, respectively. Finally, among the four machine learning algorithms
used in the study, Artificial Neural Network is found to be the best algorithm for
enhancing the performance of agricultural commodity price prediction. We also conclude
from our experiment result that considering commodity properties such as periodicity,
volatility, linearity, momentum, volume, and trend would improve the performance of
agricultural commodity price prediction. To see which of the features contributed more
to the improvement of agricultural commodity price prediction, we computed feature importance
using Random forest algorithms. The result shows that: close, high, low, open,
exponential moving average (EMA), double exponential moving average (DEMA), simple
moving average (SMA), truehigh, truelow, trend, seasonality, relative strength index
(RSI) are the most important features in sesame and coffee price prediction.
Description
Keywords
technical indicator, time series feature, price forecasting, agricultural commodity