Md. Rafiul Hassan

A fusion model of HMM, ANN and GA for stock market forecasting

In this paper we propose and implement a fusion model by combining the Hidden Markov Model (HMM), Artificial Neural Networks (ANN) and Genetic Algorithms (GA) to forecast financial market behaviour. The developed tool can be used for in depth analysis of the stock market. Using ANN, the daily stock prices are transformed to independent sets of values that become input to HMM. We draw on GA to optimize the initial parameters of HMM. The trained HMM is used to identify and locate similar patterns in the historical data. The price differences between the matched days and the respective next day are calculated. Finally, a weighted average of the price differences of similar patterns is obtained to prepare a forecast for the required next day. Forecasts are obtained for a number of securities in the IT sector and are compared with a conventional forecast method.

A hybrid of multiobjective Evolutionary Algorithm and HMM-Fuzzy model for time series prediction

In this paper, we introduce a new hybrid of Hidden Markov Model (HMM), Fuzzy Logic and multiobjective Evolutionary Algorithm (EA) for building a fuzzy model to predict non-linear time series data. In this hybrid approach, the HMMs log-likelihood score for each data pattern is used to rank the data and fuzzy rules are generated using the ranked data. We use multiobjective EA to find a range of trade-off solutions between the number of fuzzy rules and the prediction accuracy. The model is tested on a number of benchmark and more recent financial time series data. The experimental results clearly demonstrate that our model is able to generate a reduced number of fuzzy rules with similar (and in some cases better) performance compared with typical data driven fuzzy models reported in the literature.

A HMM-based adaptive fuzzy inference system for stock market forecasting

In this paper, we propose a new type of adaptive fuzzy inference system with a view to achieve improved performance for forecasting nonlinear time series data by dynamically adapting the fuzzy rules with arrival of new data. The structure of the fuzzy model utilized in the proposed system is developed based on the log-likelihood value of each data vector generated by a trained Hidden Markov Model. As part of its adaptation process, our system checks and computes the parameter values and generates new fuzzy rules as required, in response to new observations for obtaining better performance. In addition, it can also identify the most appropriate fuzzy rule in the system that covers the new data; and thus requires to adapt the parameters of the corresponding rule only, while keeping the rest of the model unchanged. This intelligent adaptive behavior enables our adaptive fuzzy inference system (FIS) to outperform standard FISs. We evaluate the performance of the proposed approach for forecasting stock price indices. The experimental results demonstrate that our approach can predict a number of stock indices, e.g., Dow Jones Industrial (DJI) index, NASDAQ index, Standard and Poor500 (S&P500) index and few other indices from UK (FTSE100), Germany (DAX) , Australia (AORD) and Japan (NIKKEI) stock markets, accurately compared with other existing computational and statistical methods.

Breast-Cancer identification using HMM-fuzzy approach

This paper presents an ensemble of feature selection and classification technique for classifying two types of breast lesion, benign and malignant. Features are selected based on their area under the ROC curves (AUC) which are then classified using a hybrid hidden Markov model (HMM)-fuzzy approach. HMM generated log-likelihood values are used to generate minimized fuzzy rules which are further optimized using gradient descent algorithms in order to enhance classification performance. The developed model is applied to Wisconsin breast cancer dataset to test its performance. The results indicate that a combination of selected features and the HMM-fuzzy approach can classify effectively the lesion types using only two fuzzy rules. Our experimental results also indicate that the proposed model can produce better classification accuracy when compared to most other computational tools.

Jaccard Index based availability prediction in enterprise grids

Enterprise Grid enables sharing and aggregation of a set of computing or storage resources connected by enterprise network, but the availability of the resources in this environment varies widely. Thus accurate prediction of the availability of these resources can significantly improve the performance of executing compute-intensive complex scientific and business applications in enterprise Grid environment by avoiding possible runtime failures. In this paper, we propose a Jaccard Index based prediction approach utilizing lazy learning algorithm that searches for a best match of a sequence pattern in the historical data in order to predict the availability of a particular machine in the system. We compare it against three other well known availability prediction techniques using simulation based study. The experimental results show that our Jaccard Index based prediction approach achieves better prediction accuracy with reduced computational complexity when compared to other similar techniques.

Improving k-Nearest Neighbour Classification with Distance Functions Based on Receiver Operating Characteristics

The k-nearest neighbour (k-NN) technique, due to its interpretable nature, is a simple and very intuitively appealing method to address classification problems. However, choosing an appropriate distance function for k-NN can be challenging and an inferior choice can make the classifier highly vulnerable to noise in the data. In this paper, we propose a new method for determining a good distance function for k-NN. Our method is based on consideration of the area under the Receiver Operating Characteristics (ROC) curve, which is a well known method to measure the quality of binary classifiers. It computes weights for the distance function, based on ROC properties within an appropriate neighbourhood for the instances whose distance is being computed. We experimentally compare the effect of our scheme with a number of other well-known k-NN distance metrics, as well as with a range of different classifiers. Experiments show that our method can substantially boost the classification performance of the k-NN algorithm. Furthermore, in a number of cases our technique is even able to deliver better accuracy than state-of-the-art non k-NN classifiers, such as support vector machines.

A novel scalable multi-class ROC for effective visualization and computation

This paper introduces a new cost function for evaluating the multi-class classifier. The new cost function facilitates both a way to visualize the performance (expected cost) of the multi-class classifier and a summary of the misclassification costs. This function overcomes the limitations of ROC in not being able to represent the classifier performance graphically when there are more than two classes. Here we present a new scalable method for producing a scalar measurement that is used to compare the performance of the multi-class classifier. We mathematically demonstrate that our technique can capture small variations in classifier performance.

Prediction of non-hydrocarbon gas components in separator by using Hybrid Computational Intelligence models

AbstractAccurate prediction of non-hydrocarbon (Non-HC) gas components in the gas-oil separators reduces the cost of gas and oil production in petroleum engineering. However, this task is difficult because there is no known relation among the properties of crude oil and the separators. There are studies that attempt to predict hydrocarbons (HCs) components using either Computational Intelligence (CI) techniques or conventional techniques like Equitation-of-State (EOS) and Empirical Correlation (EC). In this paper, we explore the applicability of CI techniques such as Artificial Neural Network, Support Vector Regressions, and Adaptive Neuro-Fuzzy Inference System to predict the Non-HC gas components in gas-oil separator tank. Further, we incorporate Genetic Algorithms (GA) into the Hybrid Computational Intelligence (HCI) models to enhance the accuracy of prediction. GA is used to determine the most favorable values of the tuning parameters in the CI models. The performances of the CI and HCI models are compared with the performance of the conventional techniques like EOS and EC. The experimental results show that accuracy of prediction by CI and HCI models outperform the conventional methods for N2 and H2S gas components. Furthermore, the HCI models perform better than the non-optimized CI models while predicting the Non-HC gas components.

Forecasting urban air pollution using HMM-fuzzy model

In this paper, we introduce a Computational Intelligence (CI)-based method to model an hourly air pollution forecasting system that can forecast concentrations of airborne pollutant variables. We have used a hybrid approach of Hidden Markov Model (HMM) with fuzzy logic (HMM-fuzzy) to model hourly air pollution at a location related to its traffic volume and meteorological variable. The forecasting performance of this hybrid model is compared with other common tool based on Artificial Neural Network (ANN) and other fuzzy tool where rules are extracted using subtractive clustering. This research demonstrates that the HMM-fuzzy approach is effectively able to model an hourly air pollution forecasting system.

Moisture Damage Modeling in Lime and Chemically Modified Asphalt at Nanolevel Using Ensemble Computational Intelligence

This paper measures the adhesion/cohesion force among asphalt molecules at nanoscale level using an Atomic Force Microscopy (AFM) and models the moisture damage by applying state-of-the-art Computational Intelligence (CI) techniques (e.g., artificial neural network (ANN), support vector regression (SVR), and an Adaptive Neuro Fuzzy Inference System (ANFIS)). Various combinations of lime and chemicals as well as dry and wet environments are used to produce different asphalt samples. The parameters that were varied to generate different asphalt samples and measure the corresponding adhesion/cohesion forces are percentage of antistripping agents (e.g., Lime and Unichem), AFM tips K values, and AFM tip types. The CI methods are trained to model the adhesion/cohesion forces given the variation in values of the above parameters. To achieve enhanced performance, the statistical methods such as average, weighted average, and regression of the outputs generated by the CI techniques are used. The experimental results show that, of the three individual CI methods, ANN can model moisture damage to lime- and chemically modified asphalt better than the other two CI techniques for both wet and dry conditions. Moreover, the ensemble of CI along with statistical measurement provides better accuracy than any of the individual CI techniques.