Jun- Lin

A neural network with a case based dynamic window for stock trading prediction

Stock forecasting involves complex interactions between market-influencing factors and unknown random processes. In this study, an integrated system, CBDWNN by combining dynamic time windows, case based reasoning (CBR), and neural network for stock trading prediction is developed and it includes three different stages: (1) screening out potential stocks and the important influential factors; (2) using back propagation network (BPN) to predict the buy/sell points (wave peak and wave trough) of stock price and (3) adopting case based dynamic window (CBDW) to further improve the forecasting results from BPN. The system developed in this research is a first attempt in the literature to predict the sell/buy decision points instead of stock price itself. The empirical results show that the CBDW can assist the BPN to reduce the false alarm of buying or selling decisions. Nine different stocks with different trends, i.e., upward, downward and steady, are studied and one individual stock (AUO) will be studied as case example. The rates of return for upward, steady, and downward trend stocks are higher than 93.57%, 37.75%, and 46.62%, respectively. These results are all very promising and better than using CBR or BPN alone.

Density-based microaggregation for statistical disclosure control

Protection of personal data in statistical databases has recently become a major societal concern. Statistical disclosure control (SDC) is often applied to statistical databases before they are released for public use. Microaggregation for SDC is a family of methods to protect microdata (i.e., records on individuals and/or companies) from individual identification. Microaggregation works by partitioning the microdata into groups of at least k records and, then, replacing the records in each group with the centroid of the group. An optimal microaggregation method must minimize the information loss resulting from this replacement process. However, this problem of minimizing information loss has been shown to be NP-hard for multivariate data. Methods based on various heuristics have been proposed for this problem, but none performs the best for every microdata set and various k values. This work presents a density-based algorithm (DBA) for microaggregation. The DBA first forms groups of records by the descending order of their densities, then fine-tunes these groups in reverse order. The performance of the DBA is compared against the latest microaggregation methods. Experimental results indicate that DBA incurs the least information loss for over half of the test situations.

Motor shaft misalignment detection using multiscale entropy with wavelet denoising

Misalignment of motor shaft (also manifesting as static eccentricity) is a common motor fault resulting from improper installation or damage of the machine components and their support structure. Spectrum analysis is generally used for online detection of such faults. This study presents a novel approach to discover features that distinguish the vibration signals of a normal motor from those of a misaligned one. These features are obtained from the difference of multiscale entropy of a signal, before and after the signal is denoised using wavelet transform. Experimental results show that classifiers based on these features obtain better and more stable accuracy rates than those based on frequency-related features.

Trend discovery in financial time series data using a case based fuzzy decision tree

Research highlights? A novel case based fuzzy decision tree model is developed to predict the time series behavior in the future. ? Fuzzy decision tree generated from stock database can be further applied in predicting stock prices movement. ? Experimental results for test data from S&P500 and stocks from S&P500 show convincing results. In recent years, many attempts have been applied to predict the behavior of stock prices movement. However, these attempts could not build an accurate and efficient stock trading system owing to the high dimensionality and non-stationary variations of stock price within a large historic database. To solve this problem, this paper applies fuzzy logic as a data mining process to generate decision trees from a stock database containing historical information. There are many attributes in the stock database and often it is impossible to develop a mathematical model to classify the data. This paper establishes a novel case based fuzzy decision tree model to identify the most important predicting attributes, and extract a set of fuzzy decision rules that can be used to predict the time series behavior in the future. The fuzzy decision tree generated from the stock database is then converted to fuzzy rules that can be further applied in decision-making of stock prices movement based on its current condition. To demonstrate the effectiveness of the CBFDT model, it is experimentally compared with other approaches on Standard & Poors 500 (S&P500) index and some stocks in S&P500. The overall performances of CBFDT model are very convincing thus it provides a new implication of research in dealing with financial time series data.

Genetic algorithm-based clustering approach for k-anonymization

k-Anonymity has been widely adopted as a model for protecting public released microdata from individual identification. This model requires that each record must be identical to at least k - 1 other records in the anonymized dataset with respect to a set of privacy-related attributes. Although anonymizing the original dataset to satisfy the requirement of k-anonymity is easy, the anonymized dataset must preserve as much information as possible of the original dataset. Clustering techniques have recently been successfully adapted for k-anonymization. This work proposes a novel genetic algorithm-based clustering approach for k-anonymization. The proposed approach adopts various heuristics to select genes for crossover operations. Experimental results show that this approach can further reduce the information loss caused by traditional clustering-based k-anonymization techniques.

Artificial chromosomes embedded in genetic algorithm for a chip resistor scheduling problem in minimizing the makespan

The manufacturing processes of a chip resistor are very similar to a flowshop scheduling problem only with minor details which can be modeled using some extra constraints; while permutation flowshop scheduling problems (PFSPs) have attracted much attention in the research works. Many approaches like genetic algorithms were dedicated to solve PFSPs effectively and efficiently. In this paper, a novel approach is presented by embedding artificial chromosomes into the genetic algorithm to further improve the solution quality and to accelerate the convergence rate. The artificial chromosome generation mechanism first analyzes the job and position association existed in previous chromosomes and records the information in an association matrix. An association matrix is generated according to the job and position distribution from top 50% chromosomes. Artificial chromosomes are determined by performing a roulette wheel selection according to the marginal probability distribution of each position. Two types of PFSPs are considered for evaluation. One is a three-machine flowshop in the printing operation of a real-world chip resistor factory and the other is the standard benchmark problems retrieved from OR-Library. The result indicates that the proposed method is able to improve the solution quality significantly and accelerate the convergence process.

Privacy preserving itemset mining through noisy items

This work investigates the problem of privacy-preserving mining of frequent itemsets. We propose a procedure to protect the privacy of data by adding noisy items to each transaction. Then, an algorithm is proposed to reconstruct frequent itemsets from these noise-added transactions. The experimental results indicate that this method can achieve a rather high level of accuracy. Our method utilizes existing algorithms for frequent itemset mining, and thereby takes full advantage of their progress to mine frequent itemset efficiently.

Minimizing a nonlinear function under a fuzzy max-t-norm relational equation constraint

This work studies a nonlinear optimization problem subject to fuzzy relational equations with max-t-norm composition. Since the feasible domain of fuzzy relational equations with more than one minimal solution is non-convex, traditional nonlinear programming methods usually cannot solve them efficiently. This work proposes a genetic algorithm to solve this problem. This algorithm first locates the feasible domain through the maximum solution and the minimal solutions of the fuzzy relational equations, to significantly reduce the search space. The algorithm then executes all genetic operations inside this feasible domain, and thus avoids the need to check the feasibility of each solution generated. Moreover, it uses a local search operation to fine-tune each mutated solution. Experimental results indicate that the proposed algorithm can accelerate the searching speed and find the optimal solution.

Comparison of microaggregation approaches on anonymized data quality

Microaggregation is commonly used to protect microdata from individual identification by anonymizing dataset records such that the resulting dataset (called the anonymized dataset) satisfies the k-anonymity constraint. Since this anonymizing process degrades data quality, an effective microaggregation approach must ensure the quality of the anonymized dataset so that the anonymized dataset remains useful for further analysis. Therefore, the performance of a microaggregation approach should be measured by the quality of the anonymized dataset generated by the microaggregation approach. Previous studies often refer to the quality of an anonymized dataset as information loss. This study takes a different approach. Since an anonymized dataset should support further analysis, this study first builds a classifier from the anonymized dataset, and then uses the prediction accuracy of that classifier to represent the quality of the anonymized dataset. Performance results indicate that low information loss does not necessarily translate into high prediction accuracy, and vice versa. This is particularly true when the information losses of both anonymized datsets do not differ significantly.

An ensemble of neural networks for stock trading decision making

Stock turning signals detection are very interesting subject arising in numerous financial and economic planning problems. In this paper, Ensemble Neural Network system with Intelligent Piecewise Linear Representation for stock turning points detection is presented. The Intelligent piecewise linear representation method is able to generate numerous stocks turning signals from the historic data base, then Ensemble Neural Network system will be applied to train the pattern and retrieve similar stock price patterns from historic data for training. These turning signals represent short-term and long-term trading signals for selling or buying stocks from the market which are applied to forecast the future turning points from the set of test data. Experimental results demonstrate that the hybrid system can make a significant and constant amount of profit when compared with other approaches using stock data available in the market.