دانلود رایگان مقاله لاتین پیش بینی قیمت نفت خام از سایت الزویر
عنوان فارسی مقاله:
یک رگرسیون بردار پشتیبانی ICA برای پیش بینی قیمت های نفت خام
عنوان انگلیسی مقاله:
An ICA-based support vector regression scheme for forecasting crude oil prices
سال انتشار : 2016
برای دانلود رایگان مقاله پیش بینی قیمت نفت خام اینجا کلیک نمایید.
بخشی از مقاله انگلیسی:
2. Methodology
2.1. Independent component analysis ICA was originally proposed for isolating independent source signals from linearly mixed signal (Jutten and Herault, 1991). Theoretically, it is a multivariate statistical technique to estimate independent components from the observed data by using high-order statistics (Comon, 1994). In the past two decades, many scholars have contributed to examine both theoretical and application aspects of ICA. In application, the use of ICA covers different areas such as microarray data classification (Fan et al., 2009, 2010), groundwater pumping analysis (Liu et al., 2015), power system disturbance identification (Ferreira et al., 2015), financial time series forecasting (Lu et al., 2009) and portfolio selection (Hitaj et al., 2015). Although ICA has gained popularity in different areas, its application to analyze and forecast crude oil price are still rare. The fluctuation of crude oil prices could be driven by different factors such as economic situation and extreme events. Since these factors are likely to generate different influences on the crude oil prices, it is reasonable to separate them from each other to identify the main driving forces behind oil price fluctuations as well as achieve better forecasting performance. Since the process is akin to the decomposition of mixed signal, in this paper we propose to apply ICA to analyze crude oil prices and derive independent components which are linked to different categories of influential factors. Technically, the basic ICA model can be written as X ¼ AS ð1Þ where X= (x1,x2,⋯xn) T denotes n by mobservations on mixed signal (e.g. crude oil prices), S= (s1, s2,⋯sn) denotes independent n by m unknown independent source signals (or influential factors), and A denotes a n by n mixing matrix. The purpose of ICA is to obtain the de-mixing matrix W (or A−1 ) such that y ¼ WX ð2Þ where y denotes the independent components estimated from the observed data. The computation process in ICA is implemented through setting appropriate estimation principles and solving the resulting optimization models. The estimation principles help to ensure the derived source signals as independent as possible, and three commonly used ones are maximum likelihood, nongaussianity maximization, and mutual information minimization (Hyvärinen et al., 2001). Each principle will generate a specific objective function whose optimization enables the estimation of independent components. Various algorithms may be used to solve the optimization model, among which the FastICA algorithm is a popular and effective one owing to its theoretic strengths (Hyvärinen and Oja, 2000). In this paper, we follow the mutual information minimum principle and adopt FastICA algorithm to derive the independent components from the observations on crude oil prices.
برای دانلود رایگان مقاله پیش بینی قیمت نفت خام اینجا کلیک نمایید.
کلمات کلیدی:
Advances in Computer Science, Intelligent Systems and Environment https://books.google.com/books?isbn=3642237533 David Jin, Sally Lin - 2011 - Computers 3.3 Crude Oil Price Forecast Based on Multiple Regression Linear Model By ... forecasting model [US$/toe] year high scheme medium scheme low scheme ... [PDF]Forecasting The Nominal Brent Oil Price With VARs—One - IMF https://www.imf.org/external/pubs/ft/wp/2015/wp15251.pdf by B Beckers - 2015 - Cited by 2 - Related articles We carry out an ex post assessment of popular models used to forecast oil prices and propose a host of ..... futures. However, recently developed economic vector auto-regression (VAR) models in the class of .... supply, Kilian's REA index and crude oil inventories outperforms the futures forecast and ..... estimation schemes. Hybrid Soft Computing Schemes for the Prediction of Import Demand ... https://www.hindawi.com/journals/mpe/2014/257947/ by YE Shao - 2014 - Cited by 3 - Related articles Apr 7, 2014 - Real data set of crude oil in Taiwan for the period of 1993–2010 and ... regression modeling was employed to forecast the coal, oil, and gas ... Principles of Business Forecasting - Page 286 - Google Books Result https://books.google.com/books?isbn=0324311273 Keith Ord, Robert Fildes - 2013 - Business & Economics Regression ... This model is known as the CochraneIOrcutt scheme (Kutner et al., 2005, pp. ... the causal form of the model, which now includes the lagged crude oil price, makes the model more desirable than the simple extrapolative scheme. Prediction of Oil Prices Using Bagging and Random Subspace ... https://link.springer.com/chapter/10.1007/978-3-319-08156-4_34 by LA Gabralla - 2014 - Cited by 4 - Related articles Further, two meta schemes namely Bagging and Random subspace are ... Prediction oil prices Bagging Random subspace Base regression models ... Xie, W., Yu, L., Xu, S., Wang, S.-Y.: A new method for crude oil price forecasting based on ...
