Frequent Sequential Patterns Research Articles

Similarity analysis of frequent sequential activity pattern mining

Activity pattern classification is extensively studied using multi-person single-day mobile traces. However, human mobility exhibits intra-personal variability and thus single-day activity may not fully capture one’s activity patterns. This paper creates a methodological framework to analyze similarity of activity patterns using frequent sequential pattern mining when multi-person multi-day data is available. Frequent sequential pattern mining discovers the frequently occurring ordered subsequences and is a natural approach of analyzing multi-day travel patterns. Prefix-Span algorithm is implemented to extract frequent patterns for each individual. Then similarity measures are defined to describe the extent to which two travelers’ activity patterns are alike and the regularity of how one repeats her activity patterns from day to day. Based upon the pairwise similarity values between two individuals, hierarchical clustering is conducted to divide travelers into communities. To illustrate these methodologies, 349 travelers’ 19,130 travel activity sequences are extracted from the world’s first connected vehicle testbed in Ann Arbor, Michigan. Three major clusters are identified. Coupled with demographics, these clusters are characterized as “seniors”, “working class”, and “parents”, respectively. Multinomial logistic regression is employed to model to what extent the similarity of socio-demographics can explain that of travel patterns. This work can be extended to either infer an unknown user’s demographics (or customer profiling) based on her activity patterns, or to reconstruct an unknown user’s frequent activity patterns based on her demographics and other similar travelers’ patterns.

Mining DNA Sequence Patterns with Constraints Using Hybridization of Firefly and Group Search Optimization

Abstract DNA sequence mining is essential in the study of the structure and function of the DNA sequence. A few exploration works have been published in the literature concerning sequence mining in information mining task. Similarly, in our past paper, an effective sequence mining was performed on a DNA database utilizing constraint measures and group search optimization (GSO). In that study, GSO calculation was utilized to optimize the sequence extraction process from a given DNA database. However, it is apparent that, occasionally, such an arbitrary seeking system does not accompany the optimal solution in the given time. To overcome the problem, we proposed in this work multiple constraints with hybrid firefly and GSO (HFGSO) algorithm. The complete DNA sequence mining process comprised the following three modules: (i) applying prefix span algorithm; (ii) calculating the length, width, and regular expression (RE) constraints; and (iii) optimal mining via HFGSO. First, we apply the concept of prefix span, which detects the frequent DNA sequence pattern using a prefix tree. Based on this prefix tree, length, width, and RE constraints are applied to handle restrictions. Finally, we adopt the HFGSO algorithm for the completeness of the mining result. The experimentation is carried out on the standard DNA sequence dataset, and the evaluation with DNA sequence dataset and the results show that our approach is better than the existing approach.

Predicting Long-Term Trajectories of Connected Vehicles via the Prefix-Projection Technique

The vehicle location prediction based on their spatial and temporal information is an important and difficult task in many applications. In the last few years, devices, such as connected vehicles, smart phones, GPS navigation systems, and smart home appliances, have amassed the large stores of geographic data. The task of leveraging this data by employing moving objects database techniques to predict spatio-temporal locations in an accurate and efficient fashion, comprising a complete trajectory remains an actively researched area. Existing methods for frequent sequential pattern mining tend to be limited to predicting short-term partial trajectories, at extremely high computational costs. In order to address these limitations, we designed a prefix-projection-based trajectory prediction algorithm called PrefixTP, which contains three essential phases. First, data collection, connected vehicles equipped with sensors comprise a vehicle grid and generate copious amounts of spatio-temporal data, in order to communicate and share traffic information. Second, model training, examining only the prefix subsequences, and projecting only their corresponding postfix subsequences into projected sets. Finally, trajectory matching, recursively finding postfix sequences meeting the requirement of minimum support count, and outputting the most frequent sequential pattern as the most probable trajectory. Fundamentally, PrefixTP supports three trajectory matching strategies which encompass all possibilities of prediction. Extensive experiments were conducted using real world GPS data sets, and the results show, when comparing predicted complete trajectories against partial short-term trajectories with a guarantee of real-time forecasting, that PrefixTP outperforms first-order, second-order Markov models, and Apriori-based trajectory prediction algorithm.

Mining co-occurrence and sequence patterns from cancer diagnoses in New York State.

The goal of this study is to discover disease co-occurrence and sequence patterns from large scale cancer diagnosis histories in New York State. In particular, we want to identify disparities among different patient groups. Our study will provide essential knowledge for clinical researchers to further investigate comorbidities and disease progression for improving the management of multiple diseases. We used inpatient discharge and outpatient visit records from the New York State Statewide Planning and Research Cooperative System (SPARCS) from 2011-2015. We grouped each patient’s visit history to generate diagnosis sequences for seven most popular cancer types. We performed frequent disease co-occurrence mining using the Apriori algorithm, and frequent disease sequence patterns discovery using the cSPADE algorithm. Different types of cancer demonstrated distinct patterns. Disparities of both disease co-occurrence and sequence patterns were observed from patients within different age groups. There were also considerable disparities in disease co-occurrence patterns with respect to different claim types (i.e., inpatient, outpatient, emergency department and ambulatory surgery). Disparities regarding genders were mostly found where the cancer types were gender specific. Supports of most patterns were usually higher for males than for females. Compared with secondary diagnosis codes, primary diagnosis codes can convey more stable results. Two disease sequences consisting of the same diagnoses but in different orders were usually with different supports. Our results suggest that the methods adopted can generate potentially interesting and clinically meaningful disease co-occurrence and sequence patterns, and identify disparities among various patient groups. These patterns could imply comorbidities and disease progressions.

Medication Pattern Mining Considering Unbiased Frequent Use by Doctors

Background: Many previous studies on mining prescription sequences are based only on frequency information, such as the number of prescriptions and the total number of patients issued the prescription. However, in cases where a very small number of doctors issue a prescription representative of a certain medication pattern to many patients many times, the prescribing intention of this very small number of doctors has a great influence on pattern extraction, which introduces bias into the final extracted frequent prescription sequence pattern. Objectives: We attempt to extract frequent prescription sequences from more diverse perspectives by considering factors other than frequency information to ensure highly reliable medication patterns. Methods: We propose the concept of unbiased frequent use by doctors as a factor in addition to frequency information based on the hypothesis that a prescription used by many doctors unbiasedly is a highly reliable prescription. We propose a medication pattern mining method that considers unbiased frequent use by doctors. We conducted an evaluation experiment using indicators based on clinical laboratory test results as a comparative evaluation of the existing method, which relied only on frequency, and included consideration of unbiased frequent use by doctors by the proposed method. Results: The weighted average value of the top k for two different evaluation methods is obtained. Conclusions: The study suggested that our medication pattern mining method considering unbiased frequent use by doctors is useful in certain situations such as when the clinical laboratory test value is outside of the normal value range.

An empirical study of using sequential behavior pattern mining approach to predict learning styles

The learning style of a learner is an important parameter in his learning process. Therefore, learning styles should be considered in the design, development, and implementation of e-learning environments to increase learners’ performance. Thus, it is important to be able to automatically determine learning styles of learners in an e-learning environment. In this paper, we propose a sequential pattern mining approach to extract frequent sequential behavior patterns, which can separate learners with different learning styles. In this research, in order to recognize learners’ learning styles, system uses the Myers-Briggs Type Indicator’s (MBTI). The approach has been implemented and tested in an e-learning environment and the results show that learning styles of learners can be predicted with high accuracy. We show that learners with similar learning styles have similar sequential behavior patterns in interaction with an e-learning environment. A lot of frequent sequential behavior patterns were extracted which some of them have a meaningful relation with MBTI dimensions.

A Comparative Study of Frequent and Maximal Periodic Pattern Mining Algorithms in Spatiotemporal Databases

Detecting regular and efficient cyclic models is the demanding activity for data analysts due to unstructured, vigorous and enormous raw information produced from web. Many existing approaches generate large candidate patterns in the occurrence of huge and complex databases. In this work, two novel algorithms are proposed and a comparative examination is performed by considering scalability and performance parameters. The first algorithm is, EFPMA (Extended Regular Model Detection Algorithm) used to find frequent sequential patterns from the spatiotemporal dataset and the second one is, ETMA (Enhanced Tree-based Mining Algorithm) for detecting effective cyclic models with symbolic database representation. EFPMA is an algorithm grows models from both ends (prefixes and suffixes) of detected patterns, which results in faster pattern growth because of less levels of database projection compared to existing approaches such as Prefixspan and SPADE. ETMA uses distinct notions to store and manage transactions data horizontally such as segment, sequence and individual symbols. ETMA exploits a partition-and-conquer method to find maximal patterns by using symbolic notations. Using this algorithm, we can mine cyclic models in full-series sequential patterns including subsection series also. ETMA reduces the memory consumption and makes use of the efficient symbolic operation. Furthermore, ETMA only records time-series instances dynamically, in terms of character, series and section approaches respectively. The extent of the pattern and proving efficiency of the reducing and retrieval techniques from synthetic and actual datasets is a really open & challenging mining problem. These techniques are useful in data streams, traffic risk analysis, medical diagnosis, DNA sequence Mining, Earthquake prediction applications. Extensive investigational outcomes illustrates that the algorithms outperforms well towards efficiency and scalability than ECLAT, STNR and MAFIA approaches.

A text representation model using Sequential Pattern-Growth method

Text representation is an essential task in transforming the input from text into features that can be later used for further Text Mining and Information Retrieval tasks. The commonly used text representation model is Bags-of-Words (BOW) and the N-gram model. Nevertheless, some known issues of these models, which are inaccurate semantic representation of text and high dimensionality of word size combination, should be investigated. A pattern-based model named Frequent Adjacent Sequential Pattern (FASP) is introduced to represent the text using a set of sequence adjacent words that are frequently used across the document collection. The purpose of this study is to discover the similarity of textual pattern between documents that can be later converted to a set of rules to describe the main news event. The FASP is based on the Pattern-Growth’s divide-and-conquer strategy where the main difference between FASP and the prior technique is in the Pattern Generation phase. This approach is tested against the BOW and N-gram text representation model using Malay and English language news dataset with different term weightings in the Vector Space Model (VSM). The findings demonstrate that the FASP model has a promising performance in finding similarities between documents with the average vector size reduction of 34% against the BOW and 77% against the N-gram model using the Malay dataset. Results using the English dataset is also consistent, indicating that the FASP approach is also language independent.

CoreFlow: Extracting and Visualizing Branching Patterns from Event Sequences

AbstractEvent sequence datasets with high event cardinality and long sequences are difficult to visualize and analyze. In particular, it is hard to generate a high level visual summary of paths and volume of flow. Existing approaches of mining and visualizing frequent sequential patterns look promising, but have limitations in terms of scalability, interpretability and utility. We propose CoreFlow, a technique that automatically extracts and visualizes branching patterns in event sequences. CoreFlow constructs a tree by recursively applying a three‐step procedure: rank events, divide sequences into groups, and trim sequences by the chosen event. The resulting tree contains key events as nodes, and links represent aggregated flows between key events. Based on CoreFlow, we have developed an interactive system for event sequence analysis. Our approach can compute branching patterns for millions of events in a few seconds, with improved interpretability of extracted patterns compared to previous work. We also present case studies of using the system in three different domains and discuss success and failure cases of applying CoreFlow to real‐world analytic problems. These case studies call forth future research on metrics and models to evaluate the quality of visual summaries of event sequences.

Mining top- k co-occurrence items with sequential pattern

The problem of mining top-k co-occurrence items with sequential pattern is defined.Sid-set and Sid-sets with an index are used to reduce the scanning time.Three algorithms including NAM, VAM, and VIAM for mining top-k co-occurrence items with sequential pattern are proposed.Two pruning techniques are proposed and used in VAM and VIAM algorithms to improve the processing time. Frequent sequential pattern mining has become one of the most important tasks in data mining. It has many applications, such as sequential analysis, classification, and prediction. How to generate candidates and how to control the combinatorically explosive number of intermediate subsequences are the most difficult problems. Intelligent systems such as recommender systems, expert systems, and business intelligence systems use only a few patterns, namely those that satisfy a number of defined conditions. Challenges include the mining of top-k patterns, top-rank-k patterns, closed patterns, and maximal patterns. In many cases, end users need to find itemsets that occur with a sequential pattern. Therefore, this paper proposes approaches for mining top-k co-occurrence items usually found with a sequential pattern. The Naive Approach Mining (NAM) algorithm discovers top-k co-occurrence items by directly scanning the sequence database to determine the frequency of items. The Vertical Approach Mining (VAM) algorithm is based on vertical database scanning. The Vertical with Index Approach Mining (VIAM) algorithm is based on a vertical database with index scanning. VAM and VIAM use pruning strategies to reduce the search space, thus improving performance. VAM and VIAM are especially effective in mining the co-occurrence items of a long input pattern. The three algorithms were evaluated using real-world databases. The experimental results show that these algorithms perform well, especially VAM and VIAM.

A recommendation engine for travel products based on topic sequential patterns

Travel products recommendation has become one of emerging issues in the realm of recommendation systems. The widely-used collaborative filtering algorithms are usually difficult to be used for recommending travel products due to a number of reasons, including (1) the content of travel products is very complex, (2) the user-item matrix is extremely sparse, and (3) the cold-start users are widely existing. To tackle these issues, we try to exploit Web server logs for generating recommendation, and present a novel recommendation engine (SECT for short) for travel products based on topic sequential patterns. In detail, we first extract topics from semantic description of every Web page. Then, we mine topic frequent sequential patterns and their target products to form click patterns library. At last, we propose a Markov n-gram model for matching the real-time click-stream of users with the click patterns library and thus computing recommendation scores. Experimental results on a real-world travel dataset demonstrate that the SECT prevails over the state-of-art baseline algorithms. In particular, SECT shows merits in improving the both coverage and accuracy for recommending products to cold-start users. Also, SECT is effective to recommend long tail items and outperform baseline algorithms.

Distinguishing surgical behavior by sequential pattern discovery

ObjectiveEach surgical procedure is unique due to patient’s and also surgeon’s particularities. In this study, we propose a new approach to distinguish surgical behaviors between surgical sites, levels of expertise and individual surgeons thanks to a pattern discovery method. MethodsThe developed approach aims to distinguish surgical behaviors based on shared longest frequent sequential patterns between surgical process models. To allow clustering, we propose a new metric called SLFSP. The approach is validated by comparison with a clustering method using Dynamic Time Warping as a metric to characterize the similarity between surgical process models. ResultsOur method outperformed the existing approach. It was able to make a perfect distinction between surgical sites (accuracy of 100%). We reached an accuracy superior to 90% and 85% for distinguishing levels of expertise and individual surgeons. ConclusionClustering based on shared longest frequent sequential patterns outperformed the previous study based on time analysis. SignificanceThe proposed method shows the feasibility of comparing surgical process models, not only by their duration but also by their structure of activities. Furthermore, patterns may show risky behaviors, which could be an interesting information for surgical training to prevent adverse events.

Grade-based Spatio-Temporal Sequential Pattern Mining using Support and Event Index Measures

Objectives: The knowledge on cause-effect relationships between instances of real-world entities can be gathered by extracting sequential patterns from spatio-temporal databases. The discovery of the patterns in the context of space and time is a challenging issue. The sequential pattern mining algorithms designed for traditional databases may result in the loss of spatio-temporal correlations due to the improper estimations of properties related to the time and space. The proposed work approaches the problem of designing sequential pattern mining algorithm specifically for spatio-temporal event datasets. Methods/Statistical Analysis: An algorithm is proposed which is based on frequency-based measures for mining frequent spatio-temporal sequential patterns. The spatio-temporal sequential pattern mining based on Support index and Event index algorithm proposes two new parameters support index and event index which are used to scrutinize the sequences extracted from the database. A data structure is also proposed to represent the spatio-temporal data for efficient pattern mining. Findings: The proposed algorithm generates the interesting set of frequent sequential patterns. The proposed algorithm is compared with Slicing-STS-Miner and MST-ITP and the experimental results proved that the proposed algorithm performs well with the order of two to three. Application/Improvements: The proposed algorithm uses frequency-based measures rather than density-based measures. Frequency-based measures take less computational time when compared to density-based measures. The proposed technique is suitable for extracting knowledge in the form of sequential patterns from spatio-temporal point databases.

An Efficient Parallel Method for Mining Frequent Closed Sequential Patterns

Mining frequent closed sequential pattern (FCSPs) has attracted a great deal of research attention, because it is an important task in sequences mining. In recently, many studies have focused on mining frequent closed sequential patterns because, such patterns have proved to be more efficient and compact than frequent sequential patterns. Information can be fully extracted from frequent closed sequential patterns. In this paper, we propose an efficient parallel approach called parallel dynamic bit vector frequent closed sequential patterns (pDBV-FCSP) using multi-core processor architecture for mining FCSPs from large databases. The pDBV-FCSP divides the search space to reduce the required storage space and performs closure checking of prefix sequences early to reduce execution time for mining frequent closed sequential patterns. This approach overcomes the problems of parallel mining such as overhead of communication, synchronization, and data replication. It also solves the load balance issues of the workload between the processors with a dynamic mechanism that re-distributes the work, when some processes are out of work to minimize the idle CPU time.

An Improved Algorithm Research on the PrefixSpan Based on the Server Session Constraint

When we mine long sequential pattern and discover knowledge by the PrefixSpan algorithm in Web Usage Mining (WUM).The elements and the suffix sequences are much more may cause the problem of the calculation, such as the space explosion. To further solve the problem a more effective way is that. Firstly, a server session-based server log file format is proposed. Then the improved algorithm on the PrefixSpan based on server session constraint is discussed for mining frequent Sequential patterns on the website. Finally, the validity and superiority of the method are presented by the experiment in the paper.

Discovery of deep order-preserving submatrix in DNA microarray data based on sequential pattern mining

In recent years, order-preserving submatrix (OPSM) model has been widely used in gene expression data analysis. Since it focuses on the changes between the elements rather than the real value, it shows better robustness and statistical significance among results than other models do. Currently, many OPSM algorithms are heuristic. They cannot mine all OPSMs as well as the deep OPSMs which are of biological significance in gene expression data. In this paper, an exact algorithm is proposed to find OPSMs by using frequent sequential pattern mining method. Firstly, we find out all common subsequences (ACS) between any two rows through dynamic programming. Then, we store them into a suffix tree. After that, we can get all OPSMs in this suffix tree, including deep OPSMs. Verified by the real gene data and artificially synthesised data, it is proved that our algorithm is efficient and meaningful.

Identifying Cancer Disease through Deoxyribonucleic Acid (DNA) Sequential Pattern Mining

This paper aims to propose the sequential pattern discovery method of Deoxyribonucleic Acid (DNA) sequence database in order to identify cancer disease. The DNA which is composed of amino acids of gene P53 is mutated. It effects to change of P53 formation. Sequential pattern discovery is a process of extracting data to generate knowledge about the series of events that has the sequences in a certain frequency so that it creates a pattern. PrefixSpan is to propose method to find a pattern of DNA sequence database. As a result, there are various selected patterns of DNA sequence. The pattem which has high similarity is used as biomarker to identify the breast cancer disease. The performance measure of support value average is 0.8. It means that the frequent sequence pattern is high. Another measure is confidence. All of the confidence values are 1. Then, the last performance measure is lift ratio at average more than 1. It means that the composed sequence items in the pattern has high dependency and relatedness. Futhermore, the selected patterns are applied as biomarker with accuracy as 100%.

Using gameplay data to examine learning behavior patterns in a serious game

Research has shown how open-ended serious games can facilitate students' development of specific skills and improve learning performance through problem-solving. However, understanding how students learn these complex skills in a game environment is a challenge, as much research uses typical paper-and-pencil assessments and self-reported surveys or other traditional observational and quantitative methods. The purpose of this study is to identify students' learning behavior patterns of problem-solving and explore behavior patterns of different performing groups within an open-ended serious game called Alien Rescue. To accomplish this purpose, this study intends to use gameplay data by incorporating sequential pattern mining and statistical analysis. The findings of this study confirmed the results from previous research (using ex situ data such as interviews) and at the same time provide an analytical approach to understand in-depth students' sequential behavior patterns using in situ gameplay data. This study examined the frequent sequential patterns between low- and high-performing students and showed that problem-solving strategies were different between these two performing groups. By using this integrated analytical method, we can gain a better understanding of the learning pathway of students’ performance and problem-solving strategies of students with different learning characteristics in a serious games context.

An efficient method for mining frequent sequential patterns using multi-Core processors

The problem of mining frequent sequential patterns (FSPs) has attracted a great deal of research attention. Although there are many efficient algorithms for mining FSPs, the mining time is still high, especially for large or dense datasets. Parallel processing has been widely applied to improve processing speed for various problems. Some parallel algorithms have been proposed, but most of them have problems related to synchronization and load balancing. Based on a multi-core processor architecture, this paper proposes a load-balancing parallel approach called Parallel Dynamic Bit Vector Sequential Pattern Mining (pDBV-SPM) for mining FSPs from huge datasets using the dynamic bit vector data structure for fast determining support values. In the pDBV-SPM approach, the support count is sorted in ascending order before the set of frequent 1-sequences is partitioned into parts, each of which is assigned to a task on a processor so that most of the nodes in the leftmost branches will be infrequent and thus pruned during the search; this strategy helps to better balance the search tree. Experiments are conducted to verify the effectiveness of pDBV-SPM. The experimental results show that the proposed algorithm outperforms PIB-PRISM for mining FSPs in terms of mining time and memory usage.

Prefix-projection global constraint and top-k approach for sequential pattern mining

Sequential pattern mining (SPM) is an important data mining problem with broad applications. SPM is a hard problem due to the huge number of intermediate subsequences to be considered. State of the art approaches for SPM (e.g., PrefixSpan Pei et al. 2001) are largely based on the pattern-growth approach, where for each frequent prefix subsequence, only its related suffix subsequences need to be considered, and the database is recursively projected into smaller ones. Many authors have promoted the use of constraints to focus on the most promising patterns according to the interests of the end user. The top-k SPM problem is also used to cope with the difficulty of thresholding and to control the number of solutions. State of the art methods developed for SPM and top-k SPM, though efficient, are locked into a rather rigid search strategy, and suffer from the lack of declarativity and flexibility. Indeed, adding new constraints usually amounts to changing the data-structures used in the core of the algorithm, and combining these new constraints often require new developments. Recent works (e.g. Kemmar et al. 2014; Negrevergne and Guns 2015) have investigated the use of Constraint Programming (CP) for SPM. However, despite their nice declarative aspects, all these modelings have scaling problems, due to the huge size of their constraint networks. To address this issue, we propose the Prefix-Projection global constraint, which encapsulates both the subsequence relation as well as the frequency constraint. Its filtering algorithm relies on the principle of projected databases which allows to keep in the variables domain, only values leading to a frequent pattern in the database. Prefix-Projection filtering algorithm enforces domain consistency on the variable succeeding the current frequent prefix in polynomial time. This global constraint also allows for a straightforward implementation of additional constraints such as size, item membership, regular expressions and any combination of them. Experimental results show that our approach clearly outperforms existing CP approaches and competes well with the state-of-the-art methods on large datasets for mining frequent sequential patterns, sequential patterns under various constraints, and top-k sequential patterns. Unlike existing CP methods, our approach achieves a better scalability.