High utility itemset mining using binary differential evolution: An application to customer segmentation

Association rule mining is a popular data mining task for finding relationships between values from the itemsets that co-occur frequently in a transactional database. Association rule mining has many applications but the “support-confidence” framework it depends on is inadequate for many cases. In recent years, a generalised task called high utility itemset mining (HUIM) has gained much popularity; it aims at discovering itemsets that yield a high revenue as measured by a utility function. However, when facing large data volumes, the running time of state-of-the-art HUIM algorithms often grows exponentially. In this work, we investigate parallel HUIM algorithms (PHUIM) and adapt two state-of-the-art sequential HUIM algorithms for parallel processing based on the Apache Spark in-memory data processing platform. Extensive experiments on several benchmark and synthetic datasets show that the proposed methods improve considerably the efficiency of the baseline HUIM algorithms. KeywordsHigh-utility itemsetApache sparkParalleld\(^2\)HUPEFIM

High utility itemsets mining identifies itemsets whose utility satisfies a given threshold. It allows users to quantify the usefulness or preferences of items using different values. Thus, it reflects the impact of different items. High utility itemsets mining is useful in decision-making process of many applications, such as retail marketing and Web service, since items are actually different in many aspects in real applications. However, due to the lack of "downward closure property", the cost of candidate generation of high utility itemsets mining is intolerable in terms of time and memory space. This paper presents a Two-Phase algorithm which can efficiently prune down the number of candidates and precisely obtain the complete set of high utility itemsets. The performance of our algorithm is evaluated by applying it to synthetic databases and two real-world applications. It performs very efficiently in terms of speed and memory cost on large databases composed of short transactions, which are difficult for existing high utility itemsets mining algorithms to handle. Experiments on real-world applications demonstrate the significance of high utility itemsets in business decision-making, as well as the difference between frequent itemsets and high utility itemsets.

P-FHM+: Parallel high utility itemset mining algorithm for big data processing

High Utility Item-set Mining (HUIM) is the futuristic remodel version of Frequent Item Mining (FIM). It has the ability to discover customer purchase trends in the retail market. By using that knowledge, retailers can incorporate innovative schemes (discounts, cross-marketing, seasonal sales offers,... etc) to enhance profit. Even though many HUIM algorithms are proposed to detect profitable patterns, most of them cannot be applied to all kinds of retail market data sets due to certain assumptions. The first assumption is that the items always produce a positive profit. But in reality, even though overall profit could be positive, some of the items make negative profit. The second one is they are developed for static transactional data. Those are useful to take decisions at some intervals like quarterly, half-yearly, yearly. But, to take decisions at any time by analyzing the present sales trend, it is required to process the data stream.

Traditional association rule mining has been widely studied, but this is not applicable to practical applications that must consider factors such as the unit profit of the item and the purchase quantity. High-utility itemset mining (HUIM) aims to find high-utility patterns by considering the number of items purchased and the unit profit. However, most high-utility itemset mining algorithms are designed for static databases. In real-world applications (such as market analysis and business decisions), databases are usually updated by inserting new data dynamically. Some researchers have proposed algorithms for finding high-utility itemsets in dynamically updated databases. Different from the batch processing algorithms that always process the databases from scratch, the incremental HUIM algorithms update and output high-utility itemsets in an incremental manner, thereby reducing the cost of finding high-utility itemsets. This paper provides the latest research on incremental high-utility itemset mining algorithms, including methods of storing itemsets and utilities based on tree, list, array and hash set storage structures. It also points out several important derivative algorithms and research challenges for incremental high-utility itemset mining.

In modern era, due to the broad applications of uncertain data, mining itemsets over uncertain databases has paying much more attention.Association Rule Mining (ARM) is a well known and most popular technique of Data Mining.It identifies itemsets from the dataset which appears frequently and generates association rules.This is the procedure which is followed by the traditional ARM it does not consider the utility of an itemsets.In real-world applications such as retail marketing, medical diagnosis, client segmentation etc., utility of itemsets is varied on various constraints such as based on cost, profit or revenue.Utility Mining intend to discover itemsets with their utilities by considering profit, quantity, cost or other user preferences.[22]High-utility itemset mining (HUIM) has thus emerged as an important research topic in data mining.But most HUIM algorithms only handle precise data, even though big data collected in reallife applications using experimental measurements or noisy sensors is often uncertain.High-Utility Rare Itemset (HURI) mining finds itemsets from a database which have their utility no less than a given minimum utility threshold and have their support less than a given frequency threshold.Identifying high-utility rare itemsets from a database can help in better business decision making by highlighting the rare itemsets which give high profits so that they can be marketed more to earn good profit.Koh and Rountree (2005) proposed a modified apriori inverse algorithm to generate rare itemsets of user interest.In this paper we propose an efficient algorithm named Mining High Utility Rare Itemsts over Uncertain Database (HURIU) .This novel approach uses the concept of apriori inverse over uncertain databases.This paper will also give the new version or extension of the algorithm HURI proposed by Jyothi et al.The implementation of an algorithm for the analysis is done on JDK 6.1 and referred the sample dataset presented by Lan Y.et al,2015[15] for uncertain database.

High utility itemset mining is an important extension of frequent itemset mining which considers unit profits and quantities of items as external and internal utilities, respectively. Since the utility function has not downward closure property, an overestimated value of utility is obtained using an anti-monotonic upper bound of utility function to prune the search space and improve the efficiency of high utility itemset mining methods. Transaction-weighted utilization (TWU) of itemset was the first and one of the most important functions which has been used as the anti-monotonic upper bound of utility by various algorithms. A variety of high utility itemset mining methods have attempted to tighten the utility upper bound and have exploited appropriate pruning strategies to improve mining efficiency. Although TWU and its improved alternatives have attempted to increase the efficiency of high utility itemset mining methods by pruning their search spaces, they suffer from a significant number of generated candidates which are high-TWU but are not high utility itemsets. Calculating the actual utilities of low utility candidates needs to multiple scanning of the dataset and thus imposes a huge overhead to the mining methods, which can cause to lose the pruning benefits of the upper bounds. Proposing appropriate pruning strategies, exploiting efficient data structures, and using tight anti-monotonic upper bounds can overcome this problem and lead to significant performance improvement in high utility itemset mining methods. In this paper, a new projection-based method, called MAHI (matrix-aided high utility itemset mining), is introduced which uses a novel utility matrix-based pruning strategy, called MA-prune to improve the high utility itemset mining performance in terms of execution time. The experimental results show that MAHI is faster than former algorithms.

In this paper, fuzzy-set theory is first used and a new algorithm called efficient fuzzy high-utility itemset mining (EFUPM) algorithm is designed to discover the fuzzy high-utility patterns from a single machine. Two upper-bounds are then estimated to early prune the unpromising candidates in the search space. To handle the large-scale of big datasets, the Hadoop-based fuzzy high-utility pattern mining (HFUPM) algorithm is then developed to discover the fuzzy high-utility patterns based on the Hadoop framework. Experimental results show that the proposed algorithms can perform well to mine the required fuzzy high-utility patterns whether in a single machine or a large-scale environment compared to the state-of-the-art approaches.

Nature is a great source of inspiration for solving complex problems in real-world. In this paper, a hybrid nature-inspired algorithm is proposed for feature selection problem. Traditionally, the real-world datasets contain all kinds of features informative as well as non-informative. These features not only increase computational complexity of the underlying algorithm but also deteriorate its performance. Hence, there an urgent need of feature selection method that select an informative subset of features from high dimensional without compromising the performance of the underlying algorithm. In this paper, we select an informative subset of features and perform cluster analysis by employing a cross breed approach of binary particle swarm optimization (BPSO) and sine cosine algorithm (SCA) named as hybrid binary particle swarm optimization and sine cosine algorithm (HBPSOSCA). Here, we employ a V-shaped transfer function to compute the likelihood of changing position for all particles. First, the effectiveness of the proposed method is tested on ten benchmark test functions. Second, the HBPSOSCA is used for data clustering problem on seven real-life datasets taken from the UCI machine learning store and gene expression model selector. The performance of proposed method is tested in comparison to original BPSO, modified BPSO with chaotic inertia weight (C-BPSO), binary moth flame optimization algorithm, binary dragonfly algorithm, binary whale optimization algorithm, SCA, and binary artificial bee colony algorithm. The conducted analysis demonstrates that the proposed method HBPSOSCA attain better performance in comparison to the competitive methods in most of the cases.

High utility itemset mining has emerged to be an important research issue in data mining since it has a wide range of real life applications. Although a number of algorithms have been proposed in recent years, the mining efficiency is still a big challenge since these algorithms suffer from either the problem of low efficiency of calculating candidates’ utilities or the problem of generating huge number of candidates. In this paper, we propose a novel data structure named PUN-list (PU-tree-Node list), which maintains both the utility information about an itemset and utility upper bound for facilitating the processing of mining high utility itemsets. Based on PUN-lists, we present a method, named MIP (Mining high utility Itemset using PUN-Lists), for efficiently mining high utility itemsets. The efficiency of MIP is achieved with three techniques. First, itemsets are represented by a highly condensed data structure, named PUN-list, which avoids costly and repeated utility computation. Second, the utility of an itemset can be efficiently calculated by scanning the PUN-list of the itemset and the PUN-lists of long itemsets can be efficiently constructed by the PUN-lists of short itemsets. Third, by employing the utility upper bound lying in the PUN-lists as the pruning strategy, MIP directly discovers high utility itemsets from the search space, named set-enumeration tree, without generating numerous candidates. Extensive experiments on various synthetic and real datasets show that MIP is very efficient since it is much faster than HUI-Miner, d2HUP, and UP-Growth + , especially on dense datasets.

Bayesian Network is a knowledge representation formalism that has been proven to be valuable in gene regulatory network reconstruction. However, it is showed that the structure learning of Bayesian networks is an NP-hard problem. Several heuristic searching techniques have been used to find better network structures. Among these algorithms, K2 algorithm has been found to be the most successful. However, the performance of K2 algorithm is greatly affected by a prior ordering of input nodes. If using inappropriate orderings, the accuracy of the learned structures will be relatively low. This paper represents a new structure learning method to reconstruct gene networks from the observational gene expression data. The proposed method is based on Binary Particle Swarm Optimization and K2 algorithm. It firstly identifies nodes ordering from the experimental data by using Binary Particle Swarm Optimization algorithm. Then, the nodes ordering information is used as an input of K2 algorithm so as to improve the performance and efficiency of the K2 algorithm in determining the network structure. Our method is evaluated using two benchmark network datasets with known structures. Experimental studies show that the proposed method can identify networks more efficiently and accurately than comparable methods reported in the literature.

This research studies on the performance of Binary Particle Swarm Optimization (BPSO) algorithm. The first part is finding the best inertia weight of BPSO from various types of inertia weight. The second part is optimizing the cross-sectional area of steel structures and topology of bracing system under vertical and lateral load. The structures studied in the research include unbraced frames and X-braced frames. Moreover, the braced frame also investigates the influence of the classification groups of elements. The elements are classified into finer groups than the original group. The design of the structure follows the AISC code. From the investigation in the first part, a constant inertia weight of 0.98 is the best. In the second part, minimum weights of unbraced frames using BPSO are the lowest weight, except three-bays, twenty-four stories frame. For braced frames with original grouping, all examples get a lower weight than the unbraced frames. For studying the influence of group, the results of the two examples are contradictory. One bay, ten stories frame with new group has a minimum weight less than the original while three-bays but twenty-four stories frame is opposite.

Here, we suggest the possibility of optical circuit design approach by employing the binary optimization of plasmonic nano rods. The proposed mechanism is based on combination of binary particle swarm optimization (BPSO) algorithm and discrete dipole approximation method. BPSO, a group of birds including a matrix with binary entries responsible for controlling nano rods in the array, shows the presence with symbol of (‘1’) and the absence with (‘0’). The current research represents a nanoscale and compact four channels plasmonic Demultiplexer as optical circuit. It includes eight coherent perfect absorption (CPA)—type filters. The operation principle is based on the absorbable formation of a conductive path in the dielectric layer of a plasmonic nano-rods waveguide. Since the CPA efficiency depends strongly on the number of plasmonic nano-rods and the nano rods location, an efficient binary optimization method based the BPSO algorithm is used to design an optimized array of the plasmonic nano-rod in order to achieve the maximum absorption coefficient in the ‘off’ state.

Coordinated controller tuning of the boiler turbine unit is a challenging task due to the nonlinear and coupling characteristics of the system. In this paper, a new variant of binary particle swarm optimization (PSO) algorithm, called probability based binary PSO (PBPSO), is presented to tune the parameters of a coordinated controller. The simulation results show that PBPSO can effectively optimize the control parameters and achieves better control performance than those based on standard discrete binary PSO, modified binary PSO, and standard continuous PSO.

It has been seen that the power distribution in the networks is becoming more complex and heavily stressed due to the development of decentralized energy resources. In such cases, Demand Side Management (DSM) programs can be utilized in order to maintain the flexibility of the system by alternating the consumption patterns of the customers as well as controlling the loads of the main distribution network. The exploitation of artificial intelligence (AI) methods in DSM applications has been developed in recent years and Particle Swarm Optimization (PSO) is one of the highly accurate methods for resource scheduling and dispatching economically. The research is focused on optimal scheduling of residential loads using the Binary Particle Swarm Optimization (BPSO) algorithm which is the binary version of the widely used PSO algorithm and the aim of this research is to minimize the monthly electricity cost of a typical household based on the Time of Use (TOU) tariff scheme.