This paper proposes a hybrid forecasting method to forecast container throughput of Qingdao *** eliminate the influence of outliers,local outlier factor(lof) is extended to detect outliers in time series,and then diff...
详细信息
This paper proposes a hybrid forecasting method to forecast container throughput of Qingdao *** eliminate the influence of outliers,local outlier factor(lof) is extended to detect outliers in time series,and then different dummy variables are constructed to capture the effect of outliers based on domain ***,a hybrid forecasting model combining projection pursuit regression(PPR) and geneticprogramming(GP) algorithm is ***,the hybrid model is applied to forecasting container throughput of Qingdao Port and the results show that the proposed method significantly outperforms ANN,SARIMA,and PPR models.
In contribution to Dunis et al. [Modelling and Trading the Corn/Ethanol Crush Spread with Neural Networks. CIBEF Working Paper. Liverpool Business School. [GRAPHICS] ], this investigation endeavours to expand the sele...
详细信息
In contribution to Dunis et al. [Modelling and Trading the Corn/Ethanol Crush Spread with Neural Networks. CIBEF Working Paper. Liverpool Business School. [GRAPHICS] ], this investigation endeavours to expand the selection of forecasting applications by delving further into the realm of artificial intelligence and nonlinear modelling. The performances of a multilayer perceptron (MLP) neural network and higher order neural network (HONN) are gauged against a genetic programming algorithm (GPA). Further to this, a time-varying volatility filter is applied by leveraging during lower volatility regimes in order to enhance the trading performance of the spread while avoiding trading completely during times of high volatility. This paper models the corn/ethanol crush spread over a six-year period commencing on 23 March 2005 (when the ethanol futures contract was first traded on Chicago Board of Trade) through to 31 December 2010. The spread acts as a good indicator of an ethanol producer's profit margin, with corn being the principal raw ingredient used in a process called 'corn crushing' to produce ethanol as a means for alternative energy. Without leveraging, the GPA achieves the highest risk-adjusted returns followed by the HONN model. Once a time-varying leverage strategy is introduced, the ranking is maintained as GPA continues to be the most profitable model with the HONN registering the second best risk-adjusted returns, followed by the MLP neural network. On that basis, and without the benefit of hindsight as in the real world, a fund manager would have selected the GPA model regardless of whether he decides to leverage or not. Furthermore, it is also observed that the time-varying leveraging strategy significantly improves annualised returns as well as reducing maximum drawdowns, two desirable outcomes for trading and hedging.
The paper deals with symbolic regression of deterministic chaos systems using a GPA-ES system. A Lorenz attractor, Rossler attractor, Rabinovich-Fabrikant equations and a van der Pol oscillator are used as examples of...
详细信息
The paper deals with symbolic regression of deterministic chaos systems using a GPA-ES system. A Lorenz attractor, Rossler attractor, Rabinovich-Fabrikant equations and a van der Pol oscillator are used as examples of deterministic chaos systems to demonstrate significant differences in the efficiency of the symbolic regression of systems described by equations of similar complexity. Within the paper, the source of this behavior is identified in presence of structures which are hard to be discovered during the evolutionary process due to the low probability of their occurrence in the initial population and by the low chance to produce them by standard evolutionary operators given by small probability to form them in a single step and low fitness function magnitudes of inter-steps when GPA tries to form them in more steps. This low magnitude of fitness function for particular solutions tends to eliminate them, thus increasing the number of needed evolutionary steps. As the solution of identified problems, modification of terminals and related crossover and mutation operators are suggested. (c) 2013 Elsevier Ltd. All rights reserved.
This paper models and forecasts the Gold Miner Spread from 23 May 2006 to 30 June 2011. The Gold Miner Spread acts as a suitable performance indicator for the relationship between physical gold and US gold equity. The...
详细信息
This paper models and forecasts the Gold Miner Spread from 23 May 2006 to 30 June 2011. The Gold Miner Spread acts as a suitable performance indicator for the relationship between physical gold and US gold equity. The contribution of this investigation is twofold. First, the accuracy of each model is evaluated from a statistical perspective. Second, various forecasting methodologies are then applied to trade the spread. Trading models include an ARMA (12,12) model, a cointegration model, a multilayer perceptron neural network (NN), a particle swarm optimization radial basis function NN and a genetic programming algorithm (GPA). Results obtained from an out-of-sample trading simulation validate the in-sample back test as the GPA model produced the highest risk-adjusted returns. Correlation filters are also applied to enhance performance and, as a consequence, volatility is reduced by 5%, on average, while returns are improved between 2.54% and 8.11% across five of the six models. Copyright (C) 2013 John Wiley & Sons, Ltd.
This paper discusses features of multilayered evolutionary system suitable to identify various systems including their model symbolic regression. Improved sensitivity allows modeling of difficult systems as determinis...
详细信息
This paper discusses features of multilayered evolutionary system suitable to identify various systems including their model symbolic regression. Improved sensitivity allows modeling of difficult systems as deterministic chaos ones. The presented paper starts with a brief introduction to previous works and ideas which allowed to build the presented two abstraction levels system. Then the structure of genetic programming algorithm - Evolutionary Strategy hybrid system is described and analyzed, including such problems as suitability to parallel implementation, optimal set of building blocks, or initial population generating rules. GPA-ES system combines GPA to model development with ES used for model parameter estimation and optimization. Such a hybrid system eliminates many weaknesses of standard CPA. The paper concludes with examples of GPA-ES application to Lorenz and Rosler systems regression and suggests application to Neural Network Model design.
The undesired effects of data gathered from real world can be produced by the noise in the process, the bias of the sensors and the presence of hysteresis, among other uncertainty sources. Data gathered by this way ar...
详细信息
ISBN:
(纸本)9783642289309;9783642289316
The undesired effects of data gathered from real world can be produced by the noise in the process, the bias of the sensors and the presence of hysteresis, among other uncertainty sources. Data gathered by this way are called Low Quality Data (LQD). Thus, uncertainty representation tools are needed for using in learning models with this kind of data. This work presents a method to represent the uncertainty and an approach for learning white box Equation Based Models (EBM). The proficiency of the representations with different noise levels and fitness functions typology is compared. The numerical results show that the use of the described objectives improves the proficiency of the algorithms. It has been also proved that each meta-heuristic determines the typology of fitness function.
In this paper a method is proposed for constructing hierarchical fuzzy rule bases in order to model black box systems defined by input-output pairs, i.e. to solve supervised machine learning problems. The resultant hi...
详细信息
ISBN:
(纸本)9781424469208
In this paper a method is proposed for constructing hierarchical fuzzy rule bases in order to model black box systems defined by input-output pairs, i.e. to solve supervised machine learning problems. The resultant hierarchical rule base is the knowledge base, which is constructed by using structure constructing evolutionary techniques, namely, genetic and Bacterial programmingalgorithms. Applying hierarchical fuzzy rule bases is a way of reducing the complexity of the knowledge base, whereas evolutionary methods ensure a relatively efficient learning process. This is the reason of the investigation of this combination.
暂无评论