LSSVR ensemble learning with uncertain parameters for crude oil price forecasting

Yu, Lean; Xu, Huijuan; Tang, Ling

doi:10.1016/j.asoc.2016.09.023

Cited by 70 publications

(21 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this way, the problem of multi-collinearity can be tackled, which can effectively avoid interference from redundant variables, searches faster, and offers better performance when solving multi-feature prediction problems [7,38,39]. SVR, a model based on the unique theory of the structural structure risk minimization principle, can resist over-fitting and simulate non-linear relations in an effective, stable way by means of kernel function form, thus solving non-linear regression and time series problems, however, SVR is sensitive to changes in input parameters due to its inherent structure [11,40,41]. BPNN is a classic neural network model, which is developed with multi-layered perceptron.…”

Section: Oil Price Forecasting Modelmentioning

confidence: 99%

Forecasting Oil Price Using Web-based Sentiment Analysis

et al. 2019

View full text Add to dashboard Cite

International oil price forecasting is a complex and important issue in the research area of energy economy. In this paper, a new model based on web-based sentiment analysis is proposed. For the oil market, sentiment analysis is used to extract key information from web texts from the four perspectives of: compound, negative, neutral, and positive sentiment. These are constructed as feature and input into oil price forecasting models with oil price itself. Finally, we analyze the effect in various views and get some interesting discoveries. The results show that the root mean squared error can be reduced by about 0.2 and the error variance by 0.2, which means that the accuracy and stability are thereby improved. Furthermore, we find that different types of sentiments can all improve performance but by similar amounts. Last but not least, text with strong intensity can better support oil price forecasting than weaker text, for which the root mean squared error can be reduced by up to 0.5, and the number of the bad cases is reduced by 20%, indicating that text with strong intensity can correct the original oil price forecast. We believe that our research will play a strong supporting role in future research on using web information for oil price forecasting.

show abstract

Section: Oil Price Forecasting Modelmentioning

confidence: 99%

Forecasting Oil Price Using Web-based Sentiment Analysis

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Since it is difficult for the tree ensemble model to minimize loss function in (14) and (15) with traditional methods in Euclidean space, the model uses the additive manner [43]. It adds that improves the model and forms the new loss function as…”

Section: Xgboostmentioning

confidence: 99%

“…Most recently, Chen et al have studied forecasting crude oil prices using deep learning framework and have found that the random walk deep belief networks (RW-DBN) model outperforms the long short term memory (LSTM) and the random walk LSTM (RW-LSTM) models in terms of forecasting accuracy [11]. Other AI-methodologies, such as genetic algorithm [12], compressive sensing [13], least square support vector regression (LSSVR) [14], and cluster support vector machine (ClusterSVM) [15], were also applied to forecasting crude oil prices. Due to the extreme nonlinearity and nonstationarity, it is hard to achieve satisfactory results by forecasting the original time series directly.…”

Section: Introductionmentioning

confidence: 99%

A CEEMDAN and XGBOOST‐Based Approach to Forecast Crude Oil Prices

Zhou

Shi

et al. 2019

Complexity

View full text Add to dashboard Cite

Crude oil is one of the most important types of energy for the global economy, and hence it is very attractive to understand the movement of crude oil prices. However, the sequences of crude oil prices usually show some characteristics of nonstationarity and nonlinearity, making it very challenging for accurate forecasting crude oil prices. To cope with this issue, in this paper, we propose a novel approach that integrates complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and extreme gradient boosting (XGBOOST), so-called CEEMDAN-XGBOOST, for forecasting crude oil prices. Firstly, we use CEEMDAN to decompose the nonstationary and nonlinear sequences of crude oil prices into several intrinsic mode functions (IMFs) and one residue. Secondly, XGBOOST is used to predict each IMF and the residue individually. Finally, the corresponding prediction results of each IMF and the residue are aggregated as the final forecasting results. To demonstrate the performance of the proposed approach, we conduct extensive experiments on the West Texas Intermediate (WTI) crude oil prices. The experimental results show that the proposed CEEMDAN-XGBOOST outperforms some state-of-the-art models in terms of several evaluation metrics.

show abstract

“…LSSVR can improve the training speed of solving the problem by transforming the quadratic programming problem of SVR into solving the problem of linear equations [28]. Hence, LSSVR has demonstrated superior performances [29][30][31] and has been applied to many forecasting areas, including reliability analysis [32], pH indications prediction [33], crashworthiness optimization problems [34], financial prediction [15], foreign exchange rate forecasting [35], revenue forecasting [36], tourism demand forecasting [29], beta systematic risk forecasting [37], crude oil price forecasting [38,39], robust system identification with outliers [40], solar irradiance forecasting [41], wind speed prediction [42], and so on. Therefore, this paper tends to introduce this powerful forecasting technique of LSSVR to perform prediction for gold prices.…”

Section: Introductionmentioning

confidence: 99%

Using Market Sentiment Analysis and Genetic Algorithm-Based Least Squares Support Vector Regression to Predict Gold Prices

Yuan¹,

Lee²,

Chiu³

2020

IJCIS

View full text Add to dashboard Cite

Gold price prediction has long been a crucial and challenging research topic for gold investors. In conventional models, most scholars have used the historical gold price or economic indicators to forecast gold prices. The gold prices depend mainly on confidence in the current market. To reduce the time delay of economic indicators in this study, the daily online global gold news undergoes a text mining approach. An opinion score is generated by ascertaining the opinion polarity and words in the daily gold news. The opinion score represents the current market trends and used as an input predictor in the forecasting model. Subsequently, the least square support vector regression (LSSVR) that is optimized by the genetic algorithm (GA) is employed to train and predict the future gold price. The mean absolute percentage error (MAPE) is adopted to evaluate the model performance. This study is the first to use the opinion score through text mining as an input predictor to GA-LSSVR in forecasting gold prices. The experiment results demonstrate that the input predictor, opinion score, can improve the predicting ability of GA-LSSVR model in terms of MAPE.

show abstract

LSSVR ensemble learning with uncertain parameters for crude oil price forecasting

Cited by 70 publications

References 20 publications

Forecasting Oil Price Using Web-based Sentiment Analysis

Forecasting Oil Price Using Web-based Sentiment Analysis

A CEEMDAN and XGBOOST‐Based Approach to Forecast Crude Oil Prices

Using Market Sentiment Analysis and Genetic Algorithm-Based Least Squares Support Vector Regression to Predict Gold Prices

Contact Info

Product

Resources

About