Class Association Rule Mining Research Articles

Investigating the contributing factors of crashes on interstate bridges in Louisiana using latent class clustering and association rule mining

Drivers on long bridges on interstates often face unique challenges, such as restricted lane width, inadequate shoulders, and a lack of clear zones for safe recoveries. Studies on understanding the factors that contribute to crash severity on such high-risk sections of interstates are limited. This research study applies latent class clustering (LCC) to detect homogeneous clusters while accounting for unobserved heterogeneity in a dataset of 10,036 crashes that occurred over a 6-year period (2015–2020) on eight selected bridges. Utilizing the LCC method, the research identifies four optimal clusters in bridge crashes, characterized by attributes such as ‘4-lane,’ ‘6-lane,’ ‘single-vehicle crashes,’ and ‘unknown driver.’ The Association Rule Mining (ARM) approach is used to identify the important collective factors to visible injury (KAB – fatal, severe and moderate) and property damage only (PDO or no injury). In Cluster 1 (4-lane), KAB and PDO crashes differ in collision type and visibility conditions, with rear-end crashes linked to KAB and sideswipe crashes to PDO. Cluster 2 (6-lane) shows similar distinctions but lacks specific lighting associations for PDO. In Cluster 3 (single-vehicle crashes), KAB involves moderate traffic and low visibility, while PDO has lower speed limits and non-dry surfaces. Cluster 4 (unknown driver), despite overrepresenting hit-and-run cases, underscores challenges in injury crash data collection in high-volume mobility scenarios. The discussions of the findings on the severity factors in this study are expected to help traffic safety engineers, policymakers, and planners to identify effective safety countermeasures on major elevated sections.

Understanding risky driving behaviors among young novice drivers in Nigeria: A latent class analysis coupled with association rule mining approach

Traffic crashes are significant public health concern in Nigeria, particularly among young drivers. The study aims to explore the underlying pattern of risky driving behaviors and the associations with demographic factors among young drivers in Nigeria. A combined approach of Latent Class Analysis (LCA) and Association Rule Mining is applied to the dataset comprising responses from 684 young drivers who complete the “Behavior of Young Novice Drivers Scale” (BYND) questionnaires. The LCA identifies four distinct classes of drivers based on the risky behavior profiles: Reckless-Speedsters, Cautious Drivers, Distracted Multitaskers, and Emotion-impacted Drivers. Association rule mining further connects these driver classes to demographic and driving history variables, uncovering intriguing insights. Reckless-Speedsters predominantly consist of young males who engage in riskier driving behaviors, including exceeding speed limits and disregarding traffic rules. Conversely, Cautious Drivers, also predominantly young males, exhibit a safer driving profile marked by rule adherence and a notably lower crash rate. Distracted Multitaskers, sharing a demographic profile with Cautious Drivers, diverge significantly due to their higher crash involvement, hinting at a propensity for distracted driving practices. Lastly, Emotion-Impacted Drivers, primarily comprising young employed males, display behaviors influenced by emotions, shorter driving distances, and prior unsupervised driving experience. Most of the behaviors are attributed to inadequate traffic control, absence of traffic signs in most of the roads, preferential treatment, and lack of strict law enforcement in the country. The findings hold substantial implications for road safety interventions in Nigeria, urging targeted approaches to address the unique challenges presented by each driver class. With acknowledging the study limitations and advocating for future research in objective measures and emotion-behavior interactions, the comprehensive approach provides a robust foundation for enhancing road safety in the Nigerian context.

A novel software defect prediction approach via weighted classification based on association rule mining

Software defect prediction technology is used to assist software practitioners in effectively allocating test resources and identifying hidden defects in a timely manner. However, the prediction of defect-prone software using association rule mining algorithms is limited because of the unbalanced distribution of defect data. Furthermore, although the existing weighted association rule mining approach considers item strength, the weight calculation still relies on expert experience and lacks fine granularity. We propose a novel software defect prediction approach based on mutual information and correlation coefficient weighted class association rule mining (MCWCAR). The MCWCAR model employs a cost-sensitive strategy and generates frequent itemsets according to three mining objectives while maintaining the original item distribution: defective class rules, non-defective class rules, and feature association relationships. During the weighted frequent itemset mining process, it combines feature selection and itemset screening to determine the appropriate feature combination through mutual information weighted support. Meanwhile, the correlation coefficient is applied to accurately depict the correlation between feature items and defect classes, serving as the weight to mine class association rules. Additionally, to ensure that interestingness measures have asymmetry and effectively represent negative associations under the condition of class imbalance, the addedvalue is adopted in the filtering association rules. We conducted experiments on 27 open-source datasets and evaluated the performance differences between MCWCAR and state-of-the-art baseline classifiers. Experimental results demonstrate that the proposed algorithm significantly outperforms other baselines in terms of Balance, Gmean, MCC, and F-measure.

A statistical method for predicting quantitative variables in association rule mining

Background:Association rules encode common patterns and structures identified in datasets. They can be derived by association rule mining (ARM) algorithms. The association rules are human-readable and allow comprehensible predictions, unlike many other types of prediction algorithms. Objective:Classical ARM algorithms, like Apriori or FP-growth, cannot process interval or ratio scaled data (quantitative variables) which limits their applicability. Results:We address this restriction in classical ARM algorithms, making it possible to process quantitative variables on the right side of a rule. Our approach is based on applying the Kullback–Leibler divergence (KLD) to identify a rule which holistically considers complete data distributions instead of using only summary statistics.We demonstrate the new approach by using, among others, the example of predicting the length of stay of intensive care patients. The length of stay describes the number of days a patient spends in the intensive care unit. In addition, we further demonstrate our approach by predicting the credit score of bank customers and the contract duration of customers of a fictional telco company based on two publically available datasets. Conclusion:This paper shows a new approach for predicting quantitative variables in ARM. We demonstrate the new approach using the FP-growth algorithm.

Attribute Selection Based Genetic Network Programming for Intrusion Detection System

Intrusion detection, as a technology used to monitor abnormal behavior and maintain network security, has attracted many researchers’ attention in recent years. Thereinto, association rule mining is one of the mainstream methods to construct intrusion detection systems (IDS). However, the existing association rule algorithms face the challenges of high false positive rate and low detection rate. Meanwhile, too many rules might lead to the uncertainty increase that affects the performance of IDS. In order to tackle the above problems, a modified genetic network programming (GNP) is proposed for class association rule mining. Specifically, based on the property that node connections in the directed graph structure of GNP can be used to construct attribute associations, we propose to introduce information gain into GNP node selection. The most important attributes are thus selected, and the irrelevant attributes are removed before the rule is extracted. Moreover, not only the uncertainty among the class association rules is alleviated and also time consumption is reduced. The extracted rules can be applied to any classifier without affecting the detection performance. Experiment results based on NSL-KDD and KDDCup99 verify the performance of our proposed algorithm.

Stock market prediction using weighted inter-transaction class association rule mining and evolutionary algorithm

Evolutionary computation and data mining are two fascinating fields that have attracted many researchers. This paper proposes a new rule mining method, named genetic network programming (GNP), to solve the prediction problem using the evolutionary algorithm. Compared with the conventional association rule methods that do not consider the weight factor, the proposed algorithm provides many advantages in financial prediction, since it can discover relationships among the attributes of different transactions. Experimental results on data from the New York Exchange Market show that the new method outperforms other conventional models in terms of both accuracy and profitability, and the proposed method can establish more important and accurate rules than the conventional methods. The results confirmed the effectiveness of the proposed data mining method in financial prediction.

NEW DISCRETE CROW SEARCH ALGORITHM FOR CLASS ASSOCIATION RULE MINING

Associative Classification (AC) or Class Association Rule (CAR) mining is a very efficient method for the classification problem. It can build comprehensible classification models in the form of a list of simple IF-THEN classification rules from the available data. In this paper, we present a new, and improved discrete version of the Crow Search Algorithm (CSA) called NDCSA-CAR to mine the Class Association Rules. The goal of this article is to improve the data classification accuracy and the simplicity of classifiers. The authors applied the proposed NDCSA-CAR algorithm on eleven benchmark dataset and compared its result with traditional algorithms and recent well known rule-based classification algorithms. The experimental results show that the proposed algorithm outperformed other rule-based approaches in all evaluated criteria.

New Discrete Crow Search Algorithm for Class Association Rule Mining

Associative Classification (AC) or Class Association Rule (CAR) mining is a very efficient method for the classification problem. It can build comprehensible classification models in the form of a list of simple IF-THEN classification rules from the available data. In this paper, we present a new, and improved discrete version of the Crow Search Algorithm (CSA) called NDCSA-CAR to mine the Class Association Rules. The goal of this article is to improve the data classification accuracy and the simplicity of classifiers. The authors applied the proposed NDCSA-CAR algorithm on eleven benchmark dataset and compared its result with traditional algorithms and recent well known rule-based classification algorithms. The experimental results show that the proposed algorithm outperformed other rule-based approaches in all evaluated criteria.

Rule-Based Models for Risk Estimation and Analysis of In-hospital Mortality in Emergency and Critical Care.

We propose a novel method that uses associative classification and odds ratios to predict in-hospital mortality in emergency and critical care. Manual mortality risk scores have previously been used to assess the care needed for each patient and their need for palliative measures. Automated approaches allow providers to get a quick and objective estimation based on electronic health records. We use association rule mining to find relevant patterns in the dataset. The odds ratio is used instead of classical association rule mining metrics as a quality measure to analyze association instead of frequency. The resulting measures are used to estimate the in-hospital mortality risk. We compare two prediction models: one minimal model with socio-demographic factors that are available at the time of admission and can be provided by the patients themselves, namely gender, ethnicity, type of insurance, language, and marital status, and a full model that additionally includes clinical information like diagnoses, medication, and procedures. The method was tested and validated on MIMIC-IV, a publicly available clinical dataset. The minimal prediction model achieved an area under the receiver operating characteristic curve value of 0.69, while the full prediction model achieved a value of 0.98. The models serve different purposes. The minimal model can be used as a first risk assessment based on patient-reported information. The full model expands on this and provides an updated risk assessment each time a new variable occurs in the clinical case. In addition, the rules in the models allow us to analyze the dataset based on data-backed rules. We provide several examples of interesting rules, including rules that hint at errors in the underlying data, rules that correspond to existing epidemiological research, and rules that were previously unknown and can serve as starting points for future studies.

Quantitative class association rule-based approach to lecturer career promotion recommendation

This paper introduces an effective method for recommendation of career promotion for educators in universities. Given an educator with an activity profile currently in a certain career state, the system is able to recommend the next career state that is most suitable for the educator by learning the pattern of historical activities of other educators. The system also recommends activities and their volume that the educators should take in order to achieve the career state. Patterns of activities and their volume were obtained using the quantitative class-association rule mining method that classifies quantified activities to a next-state class. The experiment using educator career data taken from Indonesian universities produced several recommendations that were somewhat contrary to the opinions of experts about educators who can make a career leap, an indication that expert subjectivity was more dominant than a statistically more reliable recommendation system in making decisions.

An efficient algorithm for unique class association rule mining

Association rule mining is one of the main means in Knowledge discovery and Machine learning. Such kind of rules present knowledge of interrelations among items in a dataset. Class Association Rules (CARs) are a subset of association rules which are always mined using labeled datasets. Simply, a typical CAR has an itemset that is associated to a class label. Mining CARs is vital for construction of pattern or rule-based classification models and has received recently increasing research interest. In this work, a complete efficient but not exhaustive CAR mining algorithm (UniqAR) is introduced. UniqAR generates always and only 100% accurate CARs which are called unique association rules using two rule search hypothesis of Subsumption and Nonsense to find unique itemsets in order to generate the Unique CARs. Unlike alternatives of CAR mining algorithms, UniqAR mined association rules aren’t based on itemset frequency or item selectivity. It can generate both frequent and rare association rules. No preferences of support, coverage, or item participant in itemsets are required to be provided for the proposed mining process. The main contribution of this work to CARs’ state of the art is describing unique itemsets and class association rules and providing an efficient mining process for them. Unlike the other unique rule mining alternatives in the literature, the proposed novel mining process depends on a complete but not exhaustive search that employs rules inter-relations. UniqAR has been modeled with computational analysis and extended evaluation. It is shown that UniqAR can extract all unique itemsets for unique association mining with no need to setup any user preferences, template or any constraints. Moreover, it describes accurately the effects of different dataset criteria like number of attributes/features, feature values, cases, and class labels on UniqAR unique itemset extraction mining process in an efficient way that avoids a huge number of itemsets/cases comparisons. Results have shown that the proposed UniqAR algorithm is feasible and promising.

Data mining of solar panel data to analyze the effects of different metals in producing photocurrent

Solar panels are photosensitive devices that generate electric current by being exposed to radiant power. The “Solar Army” research group at FSU’s Department of Chemistry has been collecting solar panel data. These data consist of readings from plates that contain oxides of various metallic elements. Each of these plates is represented by a 6x6 grid of cells, where the leftmost and rightmost cells are known as standards -- typically containing copper and iron respectively. The original data has been preprocessed to represent the data for one of the 6x6 cells of a plate with one data record, including the type of electric lead used, the amount of photocurrent, and the amount of fifteen different metals present on the cell. The goal of this work was to analyze the effects different metals in generating photo current so that the effectiveness of the solar panels can be improved. We used the data mining tool Weka to preprocess and apply class association rules (CAR) mining to discover the combination metals that can produce different intensities of photo current. We also applied quantitative association rule mining to determine if the quantity of metals present can affect the photo current. These test results may be important not only to the Solar Army research group, but also to manufactures and scientists attempting to increase the effectiveness of solar panels.

Association analysis of Parkinson disease with vocal change characteristics using multi-objective metaheuristic optimization

Parkinson's disease (PD) is a neurodegenerative disorder that has important economic and social effects influencing the quality of patient life. Diagnosis of PD is performed in terms of certain criteria depending on the clinical symptom evaluation. However, this method may be inadequate, especially during the onset of the disease. Acoustic analysis of PD is a cost-effective, easy, and non-invasive method for early diagnosis. The mining of association rules is one of the problems in data mining that aims to find valuable and interesting associations in huge data sets. Although association analysis is very popular and useful, to the best of our knowledge, there is not any study on association analysis of PD using vocal change characteristics. Automatic mining of comprehensible, interesting, and accurate association rules in PD data sets containing huge numerical processed voice data is aimed in this study. Due to the numerical characteristics of the vocal attributes in pre-processed PD data, classical association rules mining methods cannot be efficiently applied to this problem. For this reason; MOPNAR, NICGAR, and QAR_CIP_NSGAII that are artificial intelligence-based algorithms were modeled for mining of numerical association rules in order to obtain better performances without using any pre-process for numerical data for the first time. Furthermore, the problem of association analysis of PD with vocal change characteristics was modeled as a multi-objective optimization problem considering many different complementary/contradictory metrics such as support, confidence, comprehensibility, interestingness, etc. in this study. According to the obtained multi-objective rule sets, the NICGAR outperformed in terms of average confidence, average CF, average netconf, average yulesQ, and average number of attributes.

Software defect prediction based on correlation weighted class association rule mining

Software defect prediction based on supervised learning plays a crucial role in guiding software testing for resource allocation. In particular, it is worth noticing that using associative classification with high accuracy and comprehensibility can predict defects. But owing to the imbalance data distribution inherent, it is easy to generate a large number of non-defective class association rules, but the defective class association rules are easily ignored. Furthermore, classical associative classification algorithms mainly measure the interestingness of rules by the occurrence frequency, such as support and confidence, without considering the importance of features, resulting in combinations of the insignificant frequent itemset. This promotes the generation of weighted associative classification. However, the feature weighting based on domain knowledge is subjective and unsuitable for a high dimensional dataset. Hence, we present a novel software defect prediction model based on correlation weighted class association rule mining (CWCAR). It leverages a multi-weighted supports-based framework rather than the traditional support-confidence approach to handle class imbalance and utilizes the correlation-based heuristic approach to assign feature weight. Besides, we also optimize the ranking, pruning and prediction stages based on weighted support. Results show that CWCAR is significantly superior to state-of-the-art classifiers in terms of Balance, MCC, and Gmean.

SemiSupervised Learning for Class Association Rule Mining Using Genetic Network Programming

Data mining extracts useful knowledge from big data. The extracted knowledge in data mining is often represented by association rules, and association rules can be also used for classification. However, when association rules for classification (called class association rules) are extracted, a large number of data with class labels are necessary, which requires a lot of cost of manual annotation. Therefore, this paper proposes a semisupervised learning method for rule extraction, where a small number of labeled data and a large number of unlabeled data are used to efficiently extract class association rules. In detail, this paper proposes two types of classifiers using class association rules: one is a classifier based on semisupervised learning, and the other is that based on both supervised and semisupervised learning. The second method builds several classifiers using supervised learning and semisupervised learning and the classification results of these classifiers are integrated to make the final decision. As an association rule mining method, Genetic Network Programming (GNP), which is one of the graph‐based evolutionary algorithms is used. GNP has shown distinguished classification ability in some applications; therefore, one of the other objectives of this paper is to extend the applicability of GNP to data mining based on semisupervised learning. In the experiments, the classification accuracy is evaluated using some benchmark datasets by comparing with some conventional methods. © 2020 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

Heuristics for interesting class association rule mining a colorectal cancer database

Colorectal cancer affects many people and is one of the most frequent causes of cancer-related deaths in many countries. Professionals of the Reina Sofia University Hospital have fed a database about this pathology, with 1516 patients and 126 attributes, for more than 10 years. Finding useful knowledge therein has shown to be a difficult endeavor. We present four heuristic operators and a complete methodology for searching for interesting rules that describe cases with complications and recurrences. Our proposal has shown some advantages over the well-known Apriori algorithm, for class association rule mining, and the adaptation of three representatives of associative classification. Besides, it has allowed us to identify rules with practical interest among the vast amount of trivial and sporadic associations.

Classification of Pruning Methodologies for Model Development using Data Mining Techniques

Knowledge discovery process deals with two essential data mining techniques, association and classification. Classification produces a set of large number of associative classification rules for a given observation. Pruning removes unnecessary class association rules without losing classification accuracy. These processes are very significant but at the same time very challenging. The experimental results and limitations of existing class association rules mining techniques have shown that there is a requirement to consider more pruning parameters so that the size of classifier can be further optimized. Here through this paper we are presenting a survey various strategies for class association rule pruning and study their effects that enables us to extract efficient compact and high confidence class association rule set and we have also proposed a pruning methodology

Mining class association rules on imbalanced class datasets

Mining class association rules on imbalanced class datasets

Intelligent optimization algorithms for the problem of mining numerical association rules

There are many effective approaches that have been proposed for association rules mining (ARM) on binary or discrete-valued data. However, in many real-world applications, the data usually consist of numerical values and the standard algorithms cannot work or give promising results on these datasets. In numerical ARM (NARM), it is a difficult problem to determine which attributes will be included in the rules to be discovered and which ones will be on the left of the rule and which ones on the right. It is also difficult to automatically adjusting of most relevant ranges for numerical attributes. Directly discovering the rules without generating the frequent itemsets as used in the literature as the first step of ARM accelerates the whole process without determining the metrics needed for this step. In classical ARM algorithms, generally one or two metrics are considered. However, mined rules are needed to be comprehensible, surprising, interesting, accurate, confidential, and etc. in many real-world applications. Adjusting all of these processes without the need for the metrics to be pre-determined for each dataset seems another problem. For these purposes, evolutionary intelligent optimization algorithms seem potential solution method for this complex problem. In this paper, the performance analysis of seven evolutionary algorithms and fuzzy evolutionary algorithms; namely Alatasetal, Alcalaetal, EARMGA, GAR, GENAR, Genetic Fuzzy Apriori, and Genetic Fuzzy AprioriDC for NARM problem has been performed within eleven real datasets for the first time. The obtained results have also been compared with the classical Apriori algorithm to show the efficiencies of the intelligent algorithms on NARM problem. Performances of eight algorithms in terms of support, confidence, number of mined rules, number of covered records, and time metrics have been comparatively performed with eleven real-world datasets. One of the best-mined rules obtained by each algorithm has been given and analyzed with respect to confidence, support, and lift metrics.

Pediatric population health analysis of southern and central Illinois region: A cross sectional retrospective study using association rule mining and multiple logistic regression

BackgroundSouthern Illinois University School of Medicine (SIUSOM) collects large amounts of data every day. SIUSOM and other similar healthcare systems are always looking for better ways to use the data to understand and address population level problems. The purpose of this study is to analyze the administrative dataset for pediatric patients served by Southern Illinois University School of Medicine (SIUSOM) to uncover patterns that correlate specific demographic information to diagnoses of pediatric diseases. The study uses a cross-sectional database of medical billing information for all pediatric patients served by SIUSOM between June 2013 and December 2016. The dataset consists of about 980.9K clinical visits for 65.4K unique patients and includes patient demographic identifiers such as their sex, date of birth, race, anonymous zipcode and primary and secondary insurance plan as well as the related pediatric diagnosis codes. The goal is to find unknown correlations in this database. MethodWe proposed a two step methodology to derive unknown correlations in SIUSOM administrative database. First, Class association rule mining was used as a well-established data mining method to generate hypothesis and derive associations of the form D → M, where D is diagnosis code of a pediatric disease and M is a patient demographic identifier (age,sex, anonymous zipcode, insurance plan, or race). The resulting associations were pruned and filtered using measures such as lift, odds ratio, relative risk, and confidence. The final associations were selected by a pediatric doctor based on their clinical significance. Second,each association rule in the final set was further validated and adjusted odds ratios were obtained using multiple logistic regression. ResultsSeveral associations were found correlating specific patients’ residential zip codes with the diagnosis codes for viral hepatitis carrier, exposure to communicable diseases, screening for mental and developmental disorder in childhood, history allergy to medications, disturbance of emotions specific to childhood, and acute sinusitis. In addition, the results show that African American patients are more likely to be screened for mental and developmental disorders compared to White patients for SIUSOM pediatric population (Odds Ratio (OR):3.56, 95% Confidence Interval (CI):[3.29,3.85]). ConclusionClass association rule mining is an effective method for detecting signals in a large patient administrative database and generating hypotheses which correlate patients’ demographics with diagnosis of pediatric diseases. A post processing of the hypotheses generated by this method is necessary to prune spurious associations and select a set of clinically relevant hypotheses.