Improving Customer Churn Classification with Ensemble Stacking Method

Awang, Mohd Khalid; Makhtar, Mokhairi; Udin, Norlina; Mansor, Nur Farraliza

doi:10.14569/ijacsa.2021.0121132

Cited by 12 publications

(6 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fourth, the performance of smart contract anomaly analysis in previous studies is still at a moderate level. For example, the 95% precision rate and 69% recall rate [32] can still be improved by adapting the ensemble approach to overcome the weaknesses of individual models [41][42][43]. Fifth, the problem of balancing the distribution of data labels on the data set interferes with the model's performance.…”

Section: *Author For Correspondencementioning

confidence: 99%

Anomaly detection in smart contracts based on optimal relevance hybrid features analysis in the Ethereum blockchain employing ensemble learning

2023

IJATEE

View full text Add to dashboard Cite

Blockchain 2.0 has revolutionized the domain by introducing blockchain as a decentralized application (DApp) development platform, previously recognized mainly in the cryptocurrency sphere. Consequently, the rise of DApp development has inadvertently camouflaged fraudulent activities within smart contracts, leading to substantial losses for investors. Implementing machine learning (ML) approaches can significantly enhance the efficacy of anomaly detection. However, many studies still grapple with selecting the most pertinent features to optimize anomaly detection levels. This challenge intensifies when managing the high-dimensional raw data extracted directly from the Ethereum blockchain network, which falls under the category of big data. Smart contracts, the core of blockchain that governs DApp logic, have increasingly become a haven for fraud. This study focuses on analyzing three primary characteristic components based on contract source code (operation code (opcode), application binary interface (ABI) code, and contract transaction) to develop anomaly detection models in smart contracts using an ensemble hybrid feature strategy. The approach involves two key stages: firstly, reducing the initial feature size through constant, quasi-constant, and variant validation; and secondly, identifying the most relevant feature set using the searching for uncorrelated list of variables (SULOV) method, grounded in the minimum redundancy maximum relevance (MRMR) principle. The anomaly detection model employs a voting ensemble technique, harnessing a dataset of the most pertinent features. The model's effectiveness is gauged by comparing its performance with individual models, including random forest (RF), k-nearest neighbor (KNN), decision tree (DT), linear discriminant analysis (LDA), and stochastic gradient descent (SGD). The findings indicate that the proposed model achieves superior anomaly detection levels, with a determination value measurement rate of 92.99%, outperforming individual classifiers using the 44 most relevant features while minimizing classification time. The model's efficiency is further corroborated through comparative analysis with previous studies and alternative methodologies using the same contract dataset. The proposed ensemble-based model significantly improves anomaly detection in contract source code analysis, employing a minimal and relevant set of features refined through the SULOV method.

show abstract

Section: *Author For Correspondencementioning

confidence: 99%

Anomaly detection in smart contracts based on optimal relevance hybrid features analysis in the Ethereum blockchain employing ensemble learning

2023

IJATEE

View full text Add to dashboard Cite

show abstract

“…Awang et al [18] used stacking generalization for integration classification and compared stacking ensemble with bagging and boosting algorithms for performance comparison, concluding that stacking integration is superior in application. The above literature uses the traditional Stacking model, and there is no corresponding improvement for the application field.…”

Section: Related Workmentioning

confidence: 99%

A WOA-Stacking Tobacco Leaf Grading Method Based on Multi-Heterogeneous Classifiers Ensemble

Hou

Zhang

Liu³

et al. 2023

Preprint

View full text Add to dashboard Cite

Tobacco leaf grading is an essential step for the rational use of tobacco resources and enhancing tobacco benefits. However, the existing tobacco leaf grading technology suffers from the problem of limited grade discrimination and low identification accuracy of single classifiers. To address these issues, we proposed a tobacco leaf grading method based on Whale Optimization Algorithm and multi-heterogeneous classifier Stacking ensemble (WOA-Stacking). In this paper, we utilized the differences in data observation and training principles of heterogeneous classifiers and integrate them using the Whale Optimization Algorithm (WOA) to build a stacked ensemble model. The fitness function is calculated based on the classification performance matrix and the discriminant function. The optimal combination of base classifier and meta classifier is selected from the classifier pool composed of 9 heterogeneous classifiers, and a data-driven classification model is constructed. Subsequently, hyperparameter tuning and hierarchical coupling are performed by Grid Search (GS) and K-fold Cross-Validation (K-CV). Finally, the model is evaluated and validated by applying the tobacco leaf grade dataset. The experimental results demonstrate that the WOA-Stacking model outperforms traditional single classification models, and achieves high accuracy in tobacco leaf grade classification. The model constructed in this paper can be applied as a data-driven general model for tobacco leaf grade classification in different regions, providing technical support towards industrialization of tobacco leaf grading.

show abstract

“…Stacking often accounts for a variety of weak learners [18]. Stacking architecture improves the accuracy of classification over a single classifier as it uses various ways to solve classification problems [19].…”

Section: Ensemble Stackingmentioning

confidence: 99%

Performance Comparison between Meta-classifier Algorithms for Heart Disease Classification

Zaini¹,

Awang²

2022

IJACSA

View full text Add to dashboard Cite

The rise in heart disease among the general population is alarming. This is because cardiovascular disease is the leading cause of death, and several studies have been conducted to assist cardiologists in identifying the primary cause of heart disease. The classification accuracy of single classifiers utilised in most recent studies to predict heart disease is quite low. The accuracy of classification can be enhanced by integrating the output of multiple classifiers in an ensemble technique. Even though they can deliver the best classification accuracy, the existing ensemble approaches that integrate all classifiers are quite resource-intensive. This study thus proposes a stacking ensemble that selects the optimal subset of classifiers to produce meta-classifiers. In addition, the research compares the effectiveness of several meta-classifiers to further enhance classification. There are ten types of algorithms, including logistic regression (LR), support vector classifier (SVC), random forest (RF), extra tree classifier (ETC), naïve bayes (NB), extreme gradient boosting (XGB), decision tree (DT), k-nearest neighbor (KNN), multilayer perceptron (MLP), and stochastic gradient descent (SGD) are used as a base classifier. The construction of the meta-classifier utilised three different algorithms consisting of LR, MLP, and SVC. The prediction results from the base classifier are then used as input for the stacking ensemble. The study demonstrates that using a stacking ensemble performs better than any other single algorithm in the base classifier. The meta-classifier of logistic regression yielded 90.16% results which is better than any base classifiers. In conclusion, we could assume that the ensemble stacking approach can be considered an additional means of achieving better accuracy and has improved the performance of the classification.

show abstract

Improving Customer Churn Classification with Ensemble Stacking Method

Cited by 12 publications

References 35 publications

Anomaly detection in smart contracts based on optimal relevance hybrid features analysis in the Ethereum blockchain employing ensemble learning

Anomaly detection in smart contracts based on optimal relevance hybrid features analysis in the Ethereum blockchain employing ensemble learning

A WOA-Stacking Tobacco Leaf Grading Method Based on Multi-Heterogeneous Classifiers Ensemble

Performance Comparison between Meta-classifier Algorithms for Heart Disease Classification

Contact Info

Product

Resources

About