Soybean Price Pattern Discovery Via Toeplitz Inverse Covariance-Based Clustering

Soybean Price Pattern Discovery Via Toeplitz Inverse Covariance-Based Clustering

Hua Ling Deng (Northeast Agricultural University, Harbin, China) and Yǔ Qiàn Sūn (Northeast Agricultural University, Harbin, China)
DOI: 10.4018/IJAEIS.2019100101

Abstract

The high volatility of world soybean prices has caused uncertainty and vulnerability particularly in the developing countries. The clustering of time series is a serviceable tool for discovering soybean price patterns in temporal data. However, traditional clustering method cannot represent the continuity of price data very well, nor keep a watchful eye on the correlation between factors. In this work, the authors use the Toeplitz Inverse Covariance-Based Clustering of Multivariate Time Series Data (TICC) to soybean price pattern discovery. This is a new method for multivariate time series clustering, which can simultaneously segment and cluster the time series data. Each pattern in the TICC method is defined by a Markov random field (MRF), characterizing the interdependencies between different factors of that pattern. Based on this representation, the characteristics of each pattern and the importance of each factor can be portrayed. The work provides a new way of thinking about market price prediction for agricultural products.
Article Preview
Top

Introduction

The soybean is one of the requisite grain crops in the world, and has been cultivated for more than 5,000 years. Due to its high nutritive value, amounts of full-fat soybeans are being used in the feed and food industry. In addition, since the soybean, as the vital economic crop, plays an important role in economics and trade around the world, the soybean price index has become an important indicator of China's economic activity. The instability of soybean prices will bring huge risks to farmers, governments, consumers, and other commercial entity involved in soybean market. Therefore, accurate analysis of soybean market price needs to be taken seriously.

There is a vast literature on analysis of agricultural product price. On the one hand, several studies examine the relationship between agricultural product price and other factors. For example, Harri et al. (2009) investigated the cointegration relationship between exchange rates, oil prices, and agricultural crop prices by k-th order Vector Autoregression (VAR) model. Nazlioglu et al. (2012) studied the dynamic relationship between world oil prices and twenty-four world agricultural commodity prices by panel cointegration and Granger causality methods. Ekananda et al. (2018) observed that the world soybean price and exchange rate may affect the domestic soybean prices positively and significantly in the short term by Bound Testing Cointegration method with Autoregressive Distributed Lag (ARDL) approach. On the other hand, some researchers prefer to fit time series by selecting an appropriate model to predict agricultural product price. For example, Assis et al. (2010) and Maizah et al. (2014) predicted Cocoa Bean price sequences and the prices of Malaysian crude palm oil by Generalized AutoRegressive Conditional Heteroskedasticity (GARCH) model respectively. Octavio et al. (2009) achieved a better predicted result of U.S. soybeans and Brazilian coffee prices by Threshold Autoregressive (TAR) model. The prediction methods above can provide some valuable information for decision makers. However, these methods either cannot overcome sensitivity to noise, or only forecast in a short-term. To solve noisy sensitivity and short-term forecasting, agricultural product price sequence can be divided into several subsequences, in which each subsequence belongs to some defined trend or “pattern” reoccurring in the future. Once these patterns come under observation, seemingly unordered price data can be interpreted as a few defined patterns. The process of finding pattern is referred to “pattern discovery”. Pattern discovery try to forecast the trend of agricultural product price rather than the price in short-term. Due to the noise of data has less influence on the trend prediction than the short-terming forecasting, pattern discovery can overcome sensitivity to noise and forecast the trend of agricultural product price (Gionis et al., 2003).

Pattern discovery from time series is of fundamental importance. Particularly with the development of sensors, time series has become an important class of temporal data objects and they can be easily obtained from many applications, e.g., daily temperatures, levels of pollution, human heartbeats, and prices of agricultural products. Different from the traditional discrete database, time series data are characterized by their continuity. Therefore, when they can be focused as fragments rather than as individual data points, interesting patterns can be discovered.

Complete Article List

Search this Journal:
Reset
Open Access Articles
Volume 11: 4 Issues (2020): 1 Released, 3 Forthcoming
Volume 10: 4 Issues (2019)
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 2 Issues (2012)
Volume 2: 2 Issues (2011)
Volume 1: 2 Issues (2010)
View Complete Journal Contents Listing