Journal of Data Science and Intelligent Systems

An Ensemble Stacking Algorithm to Improve Model Accuracy in Bankruptcy Prediction

Much Aziz Muslim — 2023-03-16

Bankruptcy analysis is needed to anticipate bankruptcy. Errors in predicting bankruptcy often cause bankruptcy. Machine learning with high accuracy to analyze reversal must continuously improve its accuracy. Many machine learning models have been applied to predict bankruptcy. However, model improvisation is still needed to improve prediction accuracy. We propose a combination model to improve the accuracy of bankruptcy prediction based on a genetic algorithm-support vector machine (GA-SVM) and stacking ensemble method. This study uses the Taiwanese Bankruptcy dataset from the Taiwan Economic Journal. Then we implement a synthetic minority over-sampling technique for handling imbalanced datasets. We select the best feature using GA-SVM, adopt a new strategy by stacking the classifier, and use extreme gradient boosting as a meta-learner. The results show superior accuracy obtained by the stacking model-based GA-SVM with an accuracy of 99.58%. The accuracy obtained is higher than just applying a single classifier. Thus, this study shows that the proposed method can predict bankruptcy with superior accuracy.

Received: 11 January 2023 | Revised: 8 March 2023 | Accepted: 14 March 2023

Conflicts of Interest

Much Aziz Muslim is an editorial board member of Journal of Data Science and Intelligent Systems and was not involved in the editorial review or the decision to publish this article. The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data available on request from the corresponding author upon reasonable request.

A Comparative Analysis of Feature Eliminator Methods to Improve Machine Learning Phishing Detection

Jibrilla Tanimu — 2023-12-04

This Machine-learning-based phishing detection employs statistical models and algorithms to assess and recognise phishing attacks. These algorithms can learn patterns and features that distinguish between phishing and non-phishing attacks once they are trained on vast amounts of data from both types of cases. Phishing detection systems can quickly evaluate considerable data, identify possible phishing attempts, and warn users of potential dangers. Machine-learning-based phishing detection systems have the potential to continuously improve their accuracy over time through ongoing feature refinement, iterative model evaluation, and algorithm optimization. In contrast to conventional techniques, these systems offer a more effective and efficient approach to identifying and mitigating phishing attacks. This research critically analyzes existing literature on phishing detection, aiming to identify all proposed features and determine the critical ones necessary for accurate and fast phishing attack detection. By eliminating unnecessary overhead, this research enhances our understanding of feature eliminator methods and their role in improving machine learning-based phishing detection. The findings would contribute to the development of more robust cybersecurity measures to combat phishing attacks, as well as advance the field's knowledge and application of machine learning in detecting and mitigating such threats. The study highlights the importance of feature selection and optimization in achieving accurate and efficient phishing detection, ultimately strengthening the overall security posture of organizations and individuals against phishing attacks.

Received: 15 September 2023 | Revised: 7 November 2023 | Accepted: 4 December 2023

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available in [PhishTank] at https://phishtank.org.

Measuring the Performance of Machine Learning Forecasting Models to Support Bitcoin Investment Decisions

Wipawan Buathong — 2023-03-24

This research proposed machine learning forecasting models to support bitcoin investment decisions based on bitcoin price and trade volume from 2019 to 2021. The moving average crossovers of 5, 30, and 90 daily closing prices and their variances were inputs loaded into decision tree, random forest, and extreme gradient boosting (XGBoost) techniques to forecast bitcoin investment strategies, including market trends, actions, and holding amounts. The research also measured the models' performance based on accuracy, precision, recall, F1-score, and area under the curve-receiver operating characteristics (AUC-ROC). The results indicated that the XGBoost is the most efficient model: (1) trend (0.930 accuracy, 0.930 precision, 0.930 recall, 0.929 F1-score, and 0.983 AUC-ROC); (2) action (0.985 accuracy, 0.985 precision, 0.985 recall, 0.985 F1-score, and 0.998 AUC-ROC); and (3) amount (0.987 accuracy, 0.987 precision, 0.987 recall, 0.987 F1-score, and 0.997 AUC-ROC). The random forest achieved the second most efficient model, while the decision tree provided the lowest forecasting results. Since the bitcoin investment market in 2022 is significantly different from the previous two years due to several negative factors, the research further validated the models' performance with an unseen data set comprising 275 days of bitcoin market prices from January 1 to October 2, 2022. All the models suggested that investors hold with half the investment consistent with the investment market in 2022. Furthermore, although the decision tree and XGBoost models forecasted the investment trend for most days as up, the random forest forecasted the trend as sideway, consistent with the 2022 trend.

Received: 23 January 2023 | Revised: 22 February 2023 | Accepted: 23 March 2023

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available in [The Securities and Exchange Commission] at https://www.sec.or.th/TH/Pages/WEEKLYREPORT-2564-12.aspx; in [Yahoo Finance] at https://finance.yahoo.com/.

Federated-Based Deep Reinforcement Learning (Fed-DRL) for Energy Management in a Distributive Wireless Network

Victor Kwaku Agbesi — 2023-06-19

Studies on developing future generation wireless systems are expected to support increased infrastructure development and device subscriptions with densely deployed base stations (BSs). Economically, decreasing BS energy consumption levels and achieving "greenness" remain key factors for the giant industry. Some research works have proposed deep reinforcement techniques to solve energy management (EM) issues in cellular networks. However, these techniques are inefficient in a distributive network environment and expose the devices to privacy issues. Federated learning (FL) is proven to enforce device privacy and train models distributively. Thus, this work proposes an autonomous switching mode framework for BSs based on federated-deep reinforcement learning (Fed-DRL) to address the aforementioned challenges encountered by prior studies. Specifically, we deploy multiple DRL agents to influence the decision of the BS for EM. On the other hand, to make DRL-based decisions feasible and satisfy device quality-of-service, we train the DRL agents distributively by employing the FL concept. The results show the effectiveness of our proposed framework under distributed network scenarios compared with other benchmark algorithms.

Received: 25 April 2023 | Revised: 14 June 2023 | Accepted: 19 June 2023

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data available on request from the corresponding author upon reasonable request.

A Study of the Effects of the Shape Parameter and Type of Data Points Locations on the Accuracy of the Hermite-Based Symmetric Approach Using Positive Definite Radial Kernels

Stephen Mkegh Nengem — 2023-08-23

Theoretical approximation ideas served as the driving force behind the research. one can see that the shape parameter's behavior is driven by the kind of problem and the analytical standards that are applied. the primary issue here is not just how it impacts the interpolant's accuracy but also how quickly it converges, or how quickly the error reduces as the number of data nodes rises. Hence, this article considers two globally supported and positive radial kernels and three different patterns of data point locations on the same computational domain. The research specifically studied the effects of the shape parameter and the type of data points locations on the accurate performance of an Hermite-based symmetric approach. The two-dimensional Helmholtz equation and the two-dimensional Poisson equation were used as test functions. The problems were first solved on the three different types of data point locations using linear Laguerre-Gaussians and then, the linear Matern. In each case, the graph of the error against the shape parameter was drawn to enable easy identification of the optimal value of the shape parameter. One important result indicated that, an improved accuracy cannot be achieved without the appropriate value of the shape parameter irrespective of the type of data site used.

Received: 22 June 2023 | Revised: 14 August 2023 | Accepted: 23 August 2023

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data available on request from the corresponding author upon reasonable request.

The Evolving Landscape of Oil and Gas Chemicals: Convergence of Artificial Intelligence and Chemical-Enhanced Oil Recovery in the Energy Transition Toward Sustainable Energy Systems and Net-Zero Emissions

Alireza Bigdeli — 2024-01-29

Chemical-enhanced oil recovery (EOR) is a field of study that can gain significantly from artificial intelligence (AI), addressing uncertainties such as mobility control, interfacial tension reduction, wettability alteration, and emulsifications. The primary objective of this paper is to introduce an integrated framework for AI and chemical EOR for energy harvest operations. Central emphasis is placed on the energy transition, with the aim of expediting the development of cleaner energy harvesting systems and attaining the goal of net-zero emission. To do so, we present how the energy transition is changing the manufacturing of the chemicals for EOR application. For this, the uncertainty associated with materials' design and critical role of the simulators for transferring the laboratory experiences into full-field implementations is discussed. The concept of digitalization and its impact on energy companies are highlighted. The role of digital twin in simulators integration is discussed, emphasizing how increased data access can help design more tolerant chemicals for harsh reservoir environments using real-time data. Also, we discuss how the chemical suppliers, research institutes, startups, and field operators can benefit from self-leaning and robotic laboratories for chemicals manufacturing. Moreover, this paper explores how including AI perspectives can improve our understanding of developing chemical formulations by blending hybrid capabilities. This approach contributes to making energy production more sustainable and aligning with the goal of zero emissions. A workflow is presented to demonstrate how the integration of AI and chemical EOR can be used for both hydrocarbon production and other energy transition operations, such as carbon capture, utilization and storage, hydrogen storage, and geothermal reservoirs. The outcome of this paper stands as a pioneering effort that uniquely addresses these challenges for both academia and the industry and can open many additional doors and identify topics requiring further investigations.

Received: 21 November 2023 | Revised: 22 January 2024 | Accepted: 29 January 2024

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.