Roselina Sallehuddin

HYBRID GREY RELATIONAL ARTIFICIAL NEURAL NETWORK AND AUTO REGRESSIVE INTEGRATED MOVING AVERAGE MODEL FOR FORECASTING TIME-SERIES DATA

The aim of this study is to develop a new hybrid model by combining a linear and nonlinear model for forecasting time-series data. The proposed model (GRANN_ARIMA) integrates nonlinear grey relational artificial neural network (GRANN) and a linear autoregressive integrated moving average (ARIMA) model by combining new features and grey relational analysis to select the appropriate inputs and hybridization succession. To validate the performance of the proposed model, small and large scale data sets are used. The forecasting performance is compared with several models, and these include: individual models (ARIMA, multiple regression, GRANN), several hybrid models (MARMA, MR_ANN, ARIMA_ANN), and an artificial neural network (ANN) trained using a Levenberg Marquardt algorithm. The experiments have shown that the proposed model has outperformed other models with 99.5% forecasting accuracy for small-scale data and 99.84% for large-scale data. The obtained empirical results have proven that the GRANN_ARIMA model can provide a better alternative for time-series forecasting due to its promising performance and capability in handling time-series data for both small- and large-scale data.

Hybrid Support Vector Regression and Autoregressive Integrated Moving Average Models Improved by Particle Swarm Optimization for Property Crime Rates Forecasting with Economic Indicators

Crimes forecasting is an important area in the field of criminology. Linear models, such as regression and econometric models, are commonly applied in crime forecasting. However, in real crimes data, it is common that the data consists of both linear and nonlinear components. A single model may not be sufficient to identify all the characteristics of the data. The purpose of this study is to introduce a hybrid model that combines support vector regression (SVR) and autoregressive integrated moving average (ARIMA) to be applied in crime rates forecasting. SVR is very robust with small training data and high-dimensional problem. Meanwhile, ARIMA has the ability to model several types of time series. However, the accuracy of the SVR model depends on values of its parameters, while ARIMA is not robust to be applied to small data sets. Therefore, to overcome this problem, particle swarm optimization is used to estimate the parameters of the SVR and ARIMA models. The proposed hybrid model is used to forecast the property crime rates of the United State based on economic indicators. The experimental results show that the proposed hybrid model is able to produce more accurate forecasting results as compared to the individual models.

Application of Grey Relational Analysis for Multivariate Time Series

Grey relational analysis (GRA) has been widely applied in analysing multivariate time series data (MTS). It is an alternate solution to the traditional statistical limitations. GRA is employed to search for grey relational grade (GRG) which can be used to describe the relationships between the data attributes and to determine the important factors that significantly influence some defined objectives. This paper demonstrates how GRA has been successfully used in identifying the significant factors that affect the grain crop yield in China from 1990 to 2003. The results from analysing the sample data revealed that the main factor that affects the trend of crop yield is the consumption of pesticide and chemical fertilizer and the least important factor to be considered is the agricultural labour. Thus, by properly adjusting the significant affecting factors, the Chinas crop yield performance can be further improved. Furthermore, GRA can provide a ranking scheme that gives the order of the grey relationship among the dependent and independent factors which leads to essential information such as which input factor need to be considered to forecast grain crop yield more precisely when using artificial neural network (ANN). In order to evaluate the performance of GRA in ANN model, a comparison is made using multiple linear regression (MR) statistical method. The results from the experiment show that ANN using GRA has outperformed the MR model with 99.0% in forecasting accuracy.

Data Modeling for Kuala Lumpur Composite Index with ANFIS

Stock market transaction is one of the most popular investments activities. There are many conventional techniques being used and these include technical and fundamental analysis. Recently, AI such as ANN, GA, FL and RS are widely used by the researchers due to their ability to predict the behavior of the stock market efficiently. In this research, a comprehensive preprocessing data modeling of stock market is developed to acquire granular information that represents the behavior of the data that is to be fed to the classifier. The pre-process methodology includes splitting, scaling, normalization, feature selection, and follows by the Ten- Fold Cross Validation method as a benchmark for estimating the predictive accuracy and effectiveness of splitting and selecting the input data. Daily data of KLCI is captured and analyzed, and it is found that the movements of the Indices are unstable; hence the forecasting process becomes difficult. A Hybrid Neurofuzzy with ANFIS is suggested to predict the behavior of the Indices. Four technical indicators are chosen to analyze the data. To verify the effectiveness of the ANFIS model, two experimental have been carried out and the results show that ANFIS method is competent in forecasting the KLCI fabulously compared to ANN.

Modelling static fire hazard in a semi-arid region using frequency analysis

Various fire hazard rating systems have been used by many countries at strategic and tactical levels for fire prevention and fire safety programs. Assigning subjective weight to parameters that cause fire hazard has been widely used to model wildland fire hazard. However, these methods are sensitive to experts’ judgements because they are independent of any statistical approaches. Therefore, in the present study, we propose a wildland fire hazard method based on frequency analysis (i.e. a probability distribution model) to identify the locations of fire hazard in north-eastern Iran, which has frequent fire. The proposed methodology uses factors that do not change or change very slowly over time to identify static fire hazard areas, such as vegetation moisture, slope, aspect, elevation, distance from roads and proximity to settlements, as essential parameters. Several probability distributions are assigned to each factor to show the possibility of fire using non-linear regressions. The results show that approximately 86% of MODerate-resolution Imaging Spectroradiometer (MODIS) hot spot data are located truly in the high fire hazard areas as identified in the present study and the most significant contributing factor to fire in Golestan Province, Iran, is elevation. The present study also reveals that approximately 14% of the total study area (~20 368 km2) has a fire hazard of 66%, which can be considered very high. Therefore, this area – located mostly in the central, west and north-east regions of Golestan Province – should be considered for an effective conservation strategy of wildland fire.

Hybridization Model of Linear and Nonlinear Time Series Data for Forecasting

The aim of this paper is to propose a novel approach in hybridizing linear and nonlinear model by incorporating several new features. The intended features are multivariate information, hybridization succession alteration, and cooperative feature selection. To assess the performance of the proposed hybrid model allegedly known as Grey Relational Artificial Neural Network (GRANN_ARIMA), extensive comparisons are done with individual model (Artificial Neural Network (ANN), Autoregressive integrated Moving Average (ARIMA) and Multiple Linear Regression (MR)) and conventional hybrid model (ARIMA_ANN) with Root Mean Square Error (RMSE), Mean Absolute Deviation (MAD), Mean Absolute Percentage Error (MAPE) and Mean Square error (MSE). The experiments have shown that the proposed hybrid model has outperformed other models with 99.5% forecasting accuracy for small-scale data and 99.84% for large-scale data. The obtained empirical results have also proved that the GRANN-ARIMA is more accurate and robust due to its promising performance and capability in handling small and large scale time series data. In addition, the implementation of cooperative feature selection has assisted the forecaster to automatically determine the optimum number of input factor amid with its importantness and consequence on the generated output.

Compact classification of optimized Boolean reasoning with Particle Swarm Optimization

Conventional cut selection in Boolean reasoning BR based discretization often produces under-optimistic prime cuts. This is due to the linearity of traditional heuristics in tackling high-dimensional space problem. We proposed a flexible yet compact and holistic solution by incorporating Particle Swarm Optimization PSO into the existing framework. The first challenge is to downsize the search space such that the probability of finding the global optimum is increased. The second task is to reconstruct the present fitness function so as to improve the classification performance of the induction algorithm, which in this case, C4.5. By injecting a filtration phase prior to the cut selection and introducing a tertiary term to the fitness function, the proposed extended BR with PSO EBRPSO discretizer is developed. Based on the evaluation using four real-world datasets i.e.: Heart, Breast, Iris and Wine, it is proven that EBRPSO outperforms the existing discretizers in terms of classification accuracy as well as reduction of the decision rules.

Hybrid Artificial Neural Network with Artificial Bee Colony Algorithm for Crime Classification

Crime prevention is an important roles in police system for any country. Crime classification is one of the components in crime prevention. In this study, we proposed a hybrid crime classification model by combining Artificial Neural Network (ANN) and Artificial Bee Colony (ABC) algorithm (codename ANN-ABC). The idea is by using ABC as a learning mechanism for ANN to overcome the ANN’s local optima problem thus produce more significant results. The ANN-ABC is applied to Communities and Crime dataset to predict ’Crime Categories’. The dataset was collected from UCI machine learning repository. The result of ANN-ABC will be compare with other classification algorithms. The experiment results show that ANN-ABC outperform other algorithms and achieved 86.48% accuracy with average 7% improvement compare to other algorithms.

Moisture Prediction in Maize using Three Term Back Propagation Neural Network

The percentage of moisture content is one of the most important indexes in maize quality evaluation. Maize with high moisture content will not stay for extended periods; hence it is important to have accurate prediction of moisture content. As aflatoxin contamination in maize is of major concern, the necessity for suitable methods to predict moisture content with less time and higher accuracy assumes greater importance. Hence, Three Term Backpropagation network is proposed as a prediction tool for moisture content on maize. The new model is an improvement on Two Term Back propagation by the addition of an extra parameter, the proportional factor, which increases the convergence speed and reduces learning stalls in the conventional neural network. The experimental results are conducted using semi-annual datasets obtained from a maize thermal dryer. The results show that the proposed model outperforms Two Term Back propagation and other prediction methods like empirical correlation, analytical models (tank in series) and genetic algorithm which were used as prediction tools. Quantitatively, Three Term Back Propagation Neural Network obtained a higher precision result with a Mean Absolute Deviation (MAD) and Mean Absolute Percentage Error (MAPE) of 0.00145 and 0.00001 respectively.

Economic indicators selection for crime rates forecasting using cooperative feature selection

Features selection in multivariate forecasting model is very important to ensure that the model is accurate. The purpose of this study is to apply the Cooperative Feature Selection method for features selection. The features are economic indicators that will be used in crime rate forecasting model. The Cooperative Feature Selection combines grey relational analysis and artificial neural network to establish a cooperative model that can rank and select the significant economic indicators. Grey relational analysis is used to select the best data series to represent each economic indicator and is also used to rank the economic indicators according to its importance to the crime rate. After that, the artificial neural network is used to select the significant economic indicators for forecasting the crime rates. In this study, we used economic indicators of unemployment rate, consumer price index, gross domestic product and consumer sentiment index, as well as data rates of property crime and violent crime for the United States. Levenberg-Marquardt neural network is used in this study. From our experiments, we found that consumer price index is an important economic indicator that has a significant influence on the violent crime rate. While for property crime rate, the gross domestic product, unemployment rate and consumer price index are the influential economic indicators. The Cooperative Feature Selection is also found to produce smaller errors as compared to Multiple Linear Regression in forecasting property and violent crime rates.