Ömer Kaan Baykan

Predicting direction of stock price index movement using artificial neural networks and support vector machines

Prediction of stock price index movement is regarded as a challenging task of financial time series prediction. An accurate prediction of stock price movement may yield profits for investors. Due to the complexity of stock market data, development of efficient models for predicting is very difficult. This study attempted to develop two efficient models and compared their performances in predicting the direction of movement in the daily Istanbul Stock Exchange (ISE) National 100 Index. The models are based on two classification techniques, artificial neural networks (ANN) and support vector machines (SVM). Ten technical indicators were selected as inputs of the proposed models. Two comprehensive parameter setting experiments for both models were performed to improve their prediction performances. Experimental results showed that average performance of ANN model (75.74%) was found significantly better than that of SVM model (71.52%).

Predicting bank financial failures using neural networks, support vector machines and multivariate statistical methods: A comparative analysis in the sample of savings deposit insurance fund (SDIF) transferred banks in Turkey

Bank failures threaten the economic system as a whole. Therefore, predicting bank financial failures is crucial to prevent and/or lessen the incoming negative effects on the economic system. This is originally a classification problem to categorize banks as healthy or non-healthy ones. This study aims to apply various neural network techniques, support vector machines and multivariate statistical methods to the bank failure prediction problem in a Turkish case, and to present a comprehensive computational comparison of the classification performances of the techniques tested. Twenty financial ratios with six feature groups including capital adequacy, asset quality, management quality, earnings, liquidity and sensitivity to market risk (CAMELS) are selected as predictor variables in the study. Four different data sets with different characteristics are developed using official financial data to improve the prediction performance. Each data set is also divided into training and validation sets. In the category of neural networks, four different architectures namely multi-layer perceptron, competitive learning, self-organizing map and learning vector quantization are employed. The multivariate statistical methods; multivariate discriminant analysis, k-means cluster analysis and logistic regression analysis are tested. Experimental results are evaluated with respect to the correct accuracy performance of techniques. Results show that multi-layer perceptron and learning vector quantization can be considered as the most successful models in predicting the financial failure of banks.

A novel hybrid algorithm based on particle swarm and ant colony optimization for finding the global minimum

Abstract This paper presents a novel hybrid algorithm based on particle swarm optimization (PSO) and ant colony optimization (ACO) and called hybrid ant particle optimization algorithm (HAP) to find global minimum. In the proposed method, ACO and PSO work separately at each iteration and produce their solutions. The best solution is selected as the global best of the system and its parameters are used to select the new position of particles and ants at the next iteration. The performance of proposed method is compared with PSO and ACO on the benchmark problems and better quality results are obtained by HAP algorithm.

Short Communication: Assessment of exercise stress testing with artificial neural network in determining coronary artery disease and predicting lesion localization

The aim of this study is to show the artificial neural network (ANN) on determination of coronary artery disease existence and localization of lesion based upon exercise stress testing (EST) data. EST and coronary angiography were performed on 330 patients. The data studied acquiring 27 verifying features was normalized employing z-score method. To select training and test data, 10-fold cross-validation methods were involved and multi-layered perceptron neural network was employed for the classification. The interpretation of EST using ANN proved 91%, 73% and 65% diagnostic accuracy for the left main coronary (LMCA), left anterior descending and left circumflex coronary arteries, respectively. Besides, 69% for the right coronary artery is also predicted. For the LMCA, a 94% negative predictive value (NPV) was obtained. This high percentage of NPV encourages the elimination of LMCA lesions. Some knowledge can also be obtained about lesion localization, besides diagnosing of coronary artery disease by the assessment of EST via ANN.

A Comparison of Artificial Intelligence Methods on Determining Coronary Artery Disease

The aim of this study is to show a comparison of multi-layered perceptron neural network (MLPNN) and support vector machine (SVM) on determination of coronary artery disease existence upon exercise stress testing (EST) data. EST and coronary angiography were performed on 480 patients with acquiring 23 verifying features from each. The robustness of the proposed methods is examined using classification accuracy, k-fold cross-validation method and Cohen’s kappa coefficient. The obtained classification accuracies are approximately 78% and 79% for MLPNN and SVM respectively. Both MLPNN and SVM methods are rather satisfactory than human-based method looking to Cohen’s kappa coefficients. Besides, SVM is slightly better than MLPNN when looking to the diagnostic accuracy, average of sensitivity and specificity, and also Cohen’s kappa coefficient.

A new feature selection method for text categorization based on information gain and particle swarm optimization

Rapid increases of the documents which are created in digital media necessitate analyze and classify of these documents automatically. Feature extraction, feature selection and classifier selection in the analysis of documents and classification affects performance. In text document categorization, it is a fundamental problem that the numbers of extracted features are a lot of. In this study, by using a new feature selection method based on IG (information gain) and PSO (particle swarm optimization) algorithms, text categorization process performed. Reuters 21.578 and Classic3 corpus were used in the experiments. The roots of the words in the texts of corpus were taken as the features. Feature selection and categorization processes performed with k-Nearest Neighbors algorithm (K-NN) and Naive Bayes classifiers by using IG and PSO algorithms. Proposed system performance was evaluated by using CA (Classification Accuracy), Precision, Recall and F-measure criteria.

Classifiers fusion in recognition of wheat varieties

Five wheat varieties (Bezostaja, Cesit1252, Daǧdas, Gerek, Kiziltan traded in Konya Exchange of Commerce, Turkey), characterized by nine geometric and three colour descriptive features have been classified by multiple classier system where pair-wise SLP or SV classifiers served as base experts. In addition to standard voting and Hastie and Tibshirani fusion rules, two new ones were suggested that allowed reducing the generalization error up to 5%. In classifying of kernel lots, we may obtain faultless grain recognition.

A new approach based on particle swarm optimization algorithm for solving data allocation problem

Abstract The effectiveness distributed database systems highly depends on the state of site that its task is to allocate fragments. This allocation purpose is performed for obtaining the minimum execute time and transaction cost of queries. There are some NP-hard problems that Data Allocation Problem (DAP) is one of them and solving this problem by means of enumeration method can be computationally expensive. Recently heuristic algorithms have been used to achieve desirable solutions. Due to fewer control parameters, robustness, speed convergence characteristics and easy adaptation to the problem, this paper propose a novel method based on Particle Swarm Optimization (PSO) algorithm which is suitable to minimize the total transmission cost for both the each site – fragment dependency and the each inter – fragment dependency. The core of the study is to solve DAP by utilizing and adaptation PSO algorithm, PSO-DAP for short. Allocation of fragments to the site has been done with PSO algorithm and its performance has been evaluated on 20 different test problems and compared with the state-of-art algorithms. Experimental results and comparisons demonstrate that proposed method generates better quality solutions in terms of execution time and total cost than compared state-of-art algorithms.

Detection of tumor with Otsu-PSO method on brain MR image

Multiple image thresholding is a popular method used to separate homogeneous subsets of gray level images. To find the optimum threshold in the image in the literature is still a research topic. Many image thresholding method uses the histogram of the image. In this study, the objective function of Otsu method which is a statistical process, Particle Swarm Optimization with an intuitive algorithm (PSO) by maximizing, the optimal threshold values on a medical image were studied to find. The values obtained were tested with a standard test image and brain magnetic resonance (MR) image exposed on the tumor region in segmentation, Otsu-PSO method performance was monitored.

Shear strength predicting of FRP-strengthened RC beams by using artificial neural networks

Abstract In this study, the efficiency of artificial neural networks (ANN) in predicting the shear strength of reinforced concrete (RC) beams, strengthened by means of externally bonded fiber-reinforced polymers (FRP), is explored. Experimental data of 96 rectangular RC beams from an existing database in the literature were used to develop the ANN model. Eight different input parameters affecting the shear strength were selected for creating the ANN structure. Each parameter was arranged in an input vector and a corresponding output vector that includes the shear strength of the RC beam. For all outputs, the ANN model was trained and tested using a three-layered back-propagation method. The initial performance of back-propagation was evaluated and discussed. In addition, the accuracy of well-known building codes in predicting the shear strength of FRP-strengthened RC beams was also examined, in a comparable way, using same test data. The study shows that the ANN model gives reasonable predictions of the ultimate shear strength of RC beams (R2≈0.93). Moreover, the study concludes that the ANN model predicts the shear strength of FRP-strengthened RC beams better than existing building code approaches.