Evolutionary Data Mining for Genomics

Evolutionary Algorithms (EAs) are stochastic search algorithms inspired by the process of neo-Darwinian evolution. The motivation for applying EAs to data mining is that they are robust, adaptive search techniques that perform a global search in the solution space. This chapter first presents a brief overview of EAs, focusing mainly on two kinds of EAs, viz. Genetic Algorithms (GAs) and Genetic Programming (GP). Then the chapter reviews the main concepts and principles used by EAs designed for solving several data mining tasks, namely: discovery of classification rules, clustering, attribute selection and attribute construction. Finally, it discusses Multi-Objective EAs, based on the concept of Pareto dominance, and their use in several data mining tasks.

Knowledge discovery and data mining have become areas of growing significance because of the recent increasing demand for KDD techniques, including those used in machine learning, databases, statistics, knowledge acquisition, data visualization, and high performance computing. Knowledge discovery and data mining can be extremely beneficial for the field of Artificial Intelligence in many areas, such as industry, commerce, government, education and so on. The relation between Knowledge and Data Mining, and Knowledge Discovery in Database (KDD) process are presented in the paper. Data mining theory, Data mining tasks, Data Mining technology and Data Mining challenges are also proposed. This is an belief abstract for an invited talk at the workshop.

In this article, we focus on evaluation techniques for data mining models after feature selection and parameter estimation have already been performed. Although performance evaluation techniques for data mining models are typically independent of the data mining task, quality measures are task specific. We discuss the rationale, and the pros and cons of different quality measures by data mining task using illustrative examples to explain the concepts. We cover the following data mining tasks: classification, regression, rank prediction, clustering, and market basket analysis. We provide detailed discussions for accuracy measures such as misclassification rate, confusion matrix, misclassification cost, quadratic and informational loss, ROC curve and lift chart, RMSE , MAD , AIC , BIC , Spearman's ρ , and Kendall's τ , and quality measures for unsupervised learning tasks, including within and between cluster distances. The following performance estimation techniques are discussed: resubstitution, holdout, v ‐fold cross‐validation, leave‐one‐out/jackknife, and bootstrap. Further, we cover statistical hypothesis testing for comparing performance of two algorithms for the same data mining task. Moreover, references for further reading on the topics covered in this article are provided.

Computational scientists must understand results from experimental, observational and computational simulation generated data to gain insights and perform knowledge discovery. As systems approach the petascale range, problems that were unimaginable a few years ago are within reach. With the increasing volume and complexity of data produced by ultra-scale simulations and high-throughput experiments, understanding the science is largely hampered by the lack of comprehensive I/O, storage, acceleration of data manipulation, analysis, and mining tools. Scientists require techniques, tools and infrastructure to facilitate better understanding of their data, in particular the ability to effectively perform complex data analysis, statistical analysis and knowledge discovery. The goal of this work is to enable more effective analysis of scientific datasets through the integration of enhancements in the I/O stack, from active storage support at the file system layer to MPI-IO and high-level I/O library layers. We propose to provide software components to accelerate data analytics, mining, I/O, and knowledge discovery for large-scale scientific applications, thereby increasing productivity of both scientists and the systems. Our approaches include 1) design the interfaces in high-level I/O libraries, such as parallel netCDF, for applications to activate data mining operations at the lower I/O layers; 2) Enhance MPI-IO runtime systems to incorporate the functionality developed as a part of the runtime system design; 3) Develop parallel data mining programs as part of runtime library for server-side file system in PVFS file system; and 4) Prototype an active storage cluster, which will utilize multicore CPUs, GPUs, and FPGAs to carry out the data mining workload.

Optimization problems arise in various fields of science, engineering, and industry. In many occasions, such optimization problems, particularly in the present scenario, involve a variety of decision variables and complex structured objectives, and constraints. Often, the classical or traditional optimization techniques face difficulty in solving such real world optimization problems in their original form. Due to deficiencies of classical optimization algorithms in solving large-scale, highly non-linear, and often non-differentiable problems, there is a need to develop efficient and robust computational algorithms, which can solve problems, numerically irrespective of their sizes. Taking inspiration from nature to develop computationally efficient algorithms is one way to deal with real world optimization problems. Broadly, one can put these algorithms in the field of computational sciences and in particular, to computational intelligence. Formally, computational intelligence (CI) is a set of nature-inspired computational methodologies and approaches to solve complex real world problems. The major constituents of CI are Fuzzy Systems (FS), Neural Networks (NN), and Swarm Intelligence (SI) and Evolutionary Computation (EC). Computational intelligence techniques are powerful, efficient, flexible, and reliable. Swarm Intelligence and Evolutionary Computation are two very useful components of computational intelligence that are primarily used to solve optimization problems. This book primarily concerns with various swarm and evolutionary optimization algorithms. This chapter provides a brief introduction to swarm and evolutionary algorithms.

A common task in data mining is the visualization of multivariate objects on scatterplots, allowing human observers to perceive subtle inter-relations in the dataset such as outliers, groupings or other regularities. Least- squares multidimensional scaling (MDS) is a well known Exploratory Data Analysis family of techniques that produce dissimilarity or distance preserving layouts in a nonlinear way. In this framework, the issue of visualizing large multidimensional datasets through MDS-based methods is addressed. An original scheme providing very accurate layouts of large datasets is introduced. It is a compromise between the computational complexity O(N5/2) and the accuracy of the solution that makes it suitable both for visualization of fairly large datasets and preprocessing in pattern recognition tasks.

Many real-world optimization problems are characterized by multiple conflicting objectives, which are known as multi-objective optimization problems (MOPs). In the last two decades, evolutionary algorithms have shown promising performance in solving various MOPs, and a large number of multi-objective evolutionary algorithms (MOEAs) have been proposed. In order to determine the most suitable MOEA for a specific MOP, it is usually necessary to perform experiments to compare the performance of multiple candidate MOEAs. In 2017, an evolutionary multi-objective optimization platform was proposed by us, called PlatEMO, which provides the source codes of many state-of-the-art MOEAs and helps researchers perform batch experiments on these MOEAs. In this work, we illustrate the method of using the newest version of PlatEMO to solve MOPs in applications, by means of a case study on the feature selection problem, which is an important and difficult task in machine learning and data mining. This paper details the method of adding the feature selection problem to PlatEMO, and presents the experimental results of eight MOEAs on nine datasets in feature selection.

Chapter 14 - Knowledge Discovery and Data Mining

In knowledge management process, data mining technique can be used to extract and discover the valuable and meaningful knowledge from a large amount of data. Nowadays, data mining has given a great deal of concern and attention in the information industry and in society as a whole. This technique is an approach that is currently receiving great attention in data analysis and it has been recognized as a newly emerging analysis tool (Osei-Bryson, 2010; Park, 2001; Sinha, 2008; Tso & Yau, 2007; Wan, 2009; Zanakis, 2005; Zhuang et al., 2009). Additionally, among the major tasks in data mining are classification and prediction; concept description; rule association; cluster analysis; outlier analysis; trend and evaluation analysis; statistical analysis and others. Classification and prediction tasks are among the popular tasks in data mining; and widely used in many areas especially for trend analysis and future planning. In fact, classification technique is supervised learning, which is the class level or prediction target is already known. As a result, the classification model which is represented through rules structures will be constructed in the classification process. In this case, the constructed model will be representing the precious knowledge and it can be used for future planning. There are many areas which adapted this approach to solve their problems such as in finance, medical, marketing, stock, telecommunication, manufacturing, health care, customer relationship and etc. However, the data mining application has not attracted much attention from people in Human Resource (HR) field (Chien & Chen, 2008; Ranjan, 2008). Besides that, in our previous study, most of the prediction applications are used to predict stock, demand, rate, risk, event and others; but there are quite limited studies on human prediction. In addition prediction applications are mainly developed in business and industrious fields; and quite restricted studies involved human talent in an organization (Jantan et al., 2009). HR data can provide a rich resource for knowledge discovery and for decision support system development. Recently, an organization has to struggle effectively in term of cost, quality, service or innovation. All these depend on having enough right people with the right skills, employed

The amount of data getting generated in any sector at present is enormous. The information flow in the pharma industry is huge. Pharma firms are progressing into increased technology-enabled products and services. Data mining, which is knowledge discovery from large sets of data, helps pharma firms to discover patterns in improving the quality of drug discovery and delivery methods. The paper aims to present how data mining is useful in the pharma industry, how its techniques can yield good results in pharma sector, and to show how data mining can really enhance in making decisions using pharmaceutical data. This conceptual paper is written based on secondary study, research and observations from magazines, reports and notes. The author has listed the types of patterns that can be discovered using data mining in pharma data. The paper shows how data mining is useful in the pharma industry and how its techniques can yield good results in pharma sector. Although much work can be produced for discovering knowledge in pharma data using data mining, the paper is limited to conceptualizing the ideas and view points at this stage; future work may include applying data mining techniques to pharma data based on primary research using the available, famous significant data mining tools. Research papers and conceptual papers related to data mining in Pharma industry are rare; this is the motivation for the paper.

A Data mining is the activity that includes the collection, the use of historical data to discover regularity, patterns and relationship in a large data set. The usefulness of data mining is to specify a pattern to be found in the data mining task. The presences of data mining is motivated by the problem of data explosion which had been experienced lately these day where many organization or company collect so many years of data (purchasing data, sales data, damage data item, transaction data, and so on). In this paper data mining methods been used to analyze the damage data of finished products, with the goal of producing a pattern of the damage product. Based on the pattern from product’s damage, can be see and the aspects that affect to the damaged product. The purpose of this study is to show information about the relationship of damage data of finished goods using C4.5 algorithms in PT. Kayu Lapis Asli Murni, and display results in the form of a decision tree mining.
 Index Terms—data mining, C4.5 algorithms, decision tree, damage data of finished products, PT. Kayu Lapis Asli Murni

Discovery of class association rules from the enormous amount of database is considered as a vital task in data mining. Nowadays, the development of computer technologies, for example, database management system and data storage has given the stage for gathering and dealing with a lot of information. Finding frequent class association rules combines two significant data mining techniques such as association rule mining and classification for maximum accuracy. Since this hybrid mining yields a large number of class rules during database scan, the processing time will increases significantly. Moreover, many of these rules may be irrelevant and sometimes it misses the most important rules during classifier construction. Rare lazy learning associative classification is the process of recognizing rare class rules without generating classifier construction by focusing on the useful features from patient data of the test instance which is important for rule generation. This method is profoundly appropriate for biomedical field like heart disease prediction system which requires higher accuracy. The proposed algorithm uses simple local caching mechanism on heart disease data and constructing classifier using cogency measure that will make the lazy associative classification fast and achieve higher accuracy than traditional algorithms. Experimental results show that the proposed cogency-based algorithm is more efficient than traditional algorithms for heart disease with high prediction accuracy and the proposed algorithm may also be useful in anomaly detection, fraud detection, detection of network failures and many more bio medical areas.

From the processing point of view, data mining is machine derivation of interesting properties (to human) from the stored data. Hence, the notion of machine oriented data modeling is explored: An attribute value, in a relational model, is a meaningful label (a property) of a set of entities (granule). A model using these granules themselves as attribute values (their bit patterns or lists of members) is called a machine oriented data model. The model provides a good database compaction and data mining environment. For moderate size databases, finding association rules, decision rules, and etc., can be reduced to easy computation of i>set theoretical operations of granules. In the second part, these notions are extended to real world objects, where the universe is granulated (clustered) into granules by binary relations. Data modeling and mining with such additional semantics are formulated and investigated. In such models, data mining is essentially a machine “calculus” of granules-granular computing.

Exploring and analyzing large datasets has become an active research area in the field of data mining in the last two decades. There had been several approaches available in the literature to investigate the large datasets that comprise of millions of data. The most important data mining approaches involved in this task are preprocessing, feature selection and classification. All the three approaches have their own importance in carrying out the task effectively. Most of the existing techniques suffer from drawbacks of high complexity and computationally costly on large data sets. Especially, the classification techniques do not provide consistent and reliable results for large datasets which makes the existing classification systems inefficient and unreliable. This study mainly focuses on develop a novel and efficient framework for analyzing and classifying a large dataset. This study proposes a novel classification approach on large datasets through the process of ensemble classification. Initially, efficient preprocessing approach based on enhanced KNN and feature selection based on genetic algorithm integrated with Kernal PCA are carried out which selects a subset of informative attributes or variables to construct models relating data. Then, Classification is carried on the selected features based on the ensemble approach to get accurate results. This research study presents two types of ensemble classifiers called homogenous and heterogeneous ensemble classifiers to evaluate the performance of the proposed system. Experimental results shows that the proposed approach provide significant results for various large datasets.

Classification is an important topic in data mining research. Given a set of data records, each of which belongs to one of a number of predefined classes, the classification problem is concerned with the discovery of classification rules that can allow records with unknown class membership to be correctly classified. Many algorithms have been developed to mine large data sets for classification models and they have been shown to be very effective. However, when it comes to determining the likelihood of each classification made, many of them are not designed with such purpose in mind. For this, they are not readily applicable to such problems as churn prediction. For such an application, the goal is not only to predict whether or not a subscriber would switch from one carrier to another, it is also important that the likelihood of the subscriber's doing so be predicted. The reason for this is that a carrier can then choose to provide a special personalized offer and services to those subscribers who are predicted with higher likelihood to churn. Given its importance, we propose a new data mining algorithm, called data mining by evolutionary learning (DMEL), to handle classification problems of which the accuracy of each predictions made has to be estimated. In performing its tasks, DMEL searches through the possible rule space using an evolutionary approach that has the following characteristics: 1) the evolutionary process begins with the generation of an initial set of first-order rules (i.e., rules with one conjunct/condition) using a probabilistic induction technique and based on these rules, rules of higher order (two or more conjuncts) are obtained iteratively; 2) when identifying interesting rules, an objective interestingness measure is used; 3) the fitness of a chromosome is defined in terms of the probability that the attribute values of a record can be correctly determined using the rules it encodes; and 4) the likelihood of predictions (or classifications) made are estimated so that subscribers can be ranked according to their likelihood to churn. Experiments with different data sets showed that DMEL is able to effectively discover interesting classification rules. In particular, it is able to predict churn accurately under different churn rates when applied to real telecom subscriber data.