A Hybrid Machine Learning Framework for Systemic Financial Crisis Detection and Dynamic Asset Allocation

In the context of household financial decisions, the biomechanical mechanism mainly refers to a series of principles in which the internal physiological processes of the body are interrelated with psychological cognition and influence the decision-making results when individuals make financial decisions. From the perspective of neurobiology, Confucian culture can influence the allocation behavior of household financial assets through the neural mechanisms in multiple regions of the brain, mainly including the prefrontal cortex, amygdala, and cingulate gyrus. From the perspective of the neuroendocrine system, the secretion levels of stress hormones such as cortisol are also closely related to household financial decisions. Stress responses, by influencing psychological cognition and neural mechanisms, have become one of the important biomechanical factors affecting financial decision-making behaviors. Confucian culture has profoundly influenced individuals’ choices in the allocation of household financial assets by affecting the triggering and cognitive assessment of stress responses, the neuroendocrine regulation related to stress responses, and exerting long-term impacts and intergenerational inheritance on stress responses, resulting in characteristics such as high savings and low participation in risky assets in the allocation of household financial assets. To verify the impact of Confucian culture on the allocation of household financial assets, this study constructs the Confucian culture variable by using the iterative principal factor method, constructs Probit and Tobit models, and conducts empirical analysis using the China Household Finance Survey data (CHFS). The study finds that Confucian culture can significantly promote savings, suppress the breadth and depth of households’ participation in the financial market, and lead to a more simplified allocation of financial assets. Heterogeneity analysis shows that its impact on elderly and low-income households is more significant. This study provides insights into the cultural and biological roots behind household financial asset allocation, offering new perspectives to explain how Confucian culture shapes financial behavior through biomechanical pathways. Future research can utilize neuroscience and genetic technologies to analyze the micro-genetic, neural regulation, and molecular connections between culture and biology in the influence of Confucian culture on financial decisions from the dimensions such as the integration of molecular genetics and cultural neuroscience, the exploration of gene-culture co-evolution, and the construction of gene expression network models, so as to contribute to the research in cultural economics.

This study proposes an investment portfolio optimization system based on the AutoML framework, which integrates Bayesian hyperparameter optimization and machine learning techniques to achieve accurate prediction of financial market volatility and dynamic asset allocation. The system first extracts multi-scale features (including short-term, medium-term, and long-term returns, volatility, trading volume, and momentum indicators) based on historical stock price data, and uses a Bayesian optimized support vector regression (SVR) model to predict the volatility for the next 20 days; At the asset allocation level, the system dynamically adjusts the weights of the investment portfolio based on the predicted volatility, controlling the target volatility at 20% while constraining the leverage ratio within the range of 0-150%. The empirical results show that: 1) The volatility prediction model has excellent performance.

We develop an analytical solution to the dynamic multi‐period portfolio choice problem of an investor with risky liabilities and time varying investment opportunities. We use the model to compare the asset allocation of investors who take liabilities into account, assuming time varying returns and a multi‐period setting with the asset allocation of myopic ALM investors. In the absence of regulatory constraints on asset allocation weights, there are significant gains to investors who have access to a dynamic asset allocation model with liabilities. The gains are smaller under the typical funding ratio constraints faced by pension funds.

Bank telemarketing campaigns often experience low subscription rates due to customer heterogeneity and severe class imbalance, which pose challenges for reliable predictive modeling. This study investigates a data-driven approach that integrates synthetic minority oversampling and cost-sensitive learning to improve the prediction of telemarketing outcomes. Experiments are conducted using the Portuguese Bank Marketing dataset, comprising 41,188 instances with a positive response rate of 11.3%. Eight machine learning models are evaluated under a unified preprocessing pipeline and five-fold stratified cross-validation, including Logistic Regression, Decision Tree, Random Forest, and Ensemble methods. The results show that Ensemble models, particularly CatBoost, XGBoost, and LightGBM, achieve improved performance compared with traditional baselines, with notable gains in minority-class recall and overall discrimination ability. The best-performing model attains an F1-score of 0.540, a recall of 0.812 for the positive class, and a ROC–AUC of 0.908. To enhance interpretability, SHAP-based analysis is applied to quantify feature contributions, identifying campaign duration, previous contact outcomes, and selected macroeconomic indicators as key predictors. These findings indicate that combining resampling strategies with cost-sensitive optimization provides a robust and transparent approach for learning from imbalanced telemarketing data, thereby supporting reproducible and data-driven financial decision-making by explicitly addressing difficulty in minority-class identification under imbalance and class imbalance under cross-entropy training in imbalanced banking data.

The global crude oil market, known for its pronounced volatility and nonlinear dynamics, plays a pivotal role in shaping economic stability and informing investment strategies. Contrary to traditional research focused on price forecasting, this study emphasizes the more investor-centric task of predicting returns for West Texas Intermediate (WTI) crude oil. By spotlighting returns, it directly addresses critical investor concerns such as asset allocation and risk management. This study applies advanced machine learning models, including XGBoost, random forest, and neural networks to predict crude oil return, and for the first time, incorporates sustainability and external risk variables, which are shown to enhance predictive performance in capturing the non-stationarity and complexity of financial time-series data. To enhance predictive accuracy, we integrate 55 variables across five dimensions: macroeconomic indicators, financial and futures markets, energy markets, momentum factors, and sustainability and external risk. Among these, the rate of change stands out as the most influential predictor. Notably, XGBoost demonstrates a superior performance, surpassing competing models with an impressive 76% accuracy in direction forecasting. The analysis highlights how the significance of various predictors shifted during the COVID-19 pandemic. This underscores the dynamic and adaptive character of crude oil markets under substantial external disruptions. In addition, by incorporating sustainability factors, the study provides deeper insights into the drivers of market behavior, supporting more informed portfolio adjustments, risk management strategies, and policy development aimed at fostering resilience and advancing sustainable energy transitions.

Strategic Asset Allocation in Japan: An Empirical Evaluation

In recent years, the demand for reliable, real-time, and energy-efficient transmission of healthcare data has grown significantly due to the proliferation of remote patient monitoring systems. The commonly used Traditional RF-based Wireless Sensor Networks (WSNs) have been characterized by electromagnetic interference, low bandwidth and excessive power consumption particularly in sensitive medical settings. To overcome these shortcomings, this study suggests a new model that integrates the Optical Wireless Communication (OWC) based on LED with Hybrid Machine Learning (ML) methods to streamline the process of transferring healthcare information. The technology incorporates Visible Light Communication (VLC) using high-efficiency LEDs and strategically placed biomedical sensor nodes in order to support a line-of-sight transmission of patient vital signs, including ECG, SpO 2 and body temperature. An ML engine with a hybrid of Reinforcement Learning (RL) which adapts routing based on the network and Random Forest (RF) which predicts link quality improves the decision-making capacity of the network. Simulations made experimentally show better performance on important measures. The model proposed has an energy consumption of 68.7mJ, a packet delivery ratio (PDR) of 98.9, a latency average of 74.3ms and a bit error rate (BER) of 0.006 which is better than the RF-based model and other intelligent WSN models.

Leveraging hybrid machine learning and data fusion for accurate mapping of malaria cases using meteorological variables in western India

Canner, Mankiw and Weil (1997) point out that the popular financial advisors on portfolio allocation among cash, bonds and stocks appear not to follow the mutual-fund separation theorem and call the inconsistency between separation theorem and popular financial advice ”an asset allocation puzzle.” For solving the asset allocation puzzle, we provide an analysis of the optimal dynamic asset allocation strategy for a long-horizon investor who has nontraded assets under an economic environment with stochastic investment opportunities and incomplete financial markets. We propose another distinguishing hedging component of the dynamic asset allocation for the stock index fund: the human capital hedging component, the hedging demand which characterizes the demand arising from the desire to hedge against changes in the labor income in contrast to Merton (1973). When we incorporate nontraded assets with intertemporal hedging, we can solve the asset allocation puzzle successfully.

Concrete is the most widely used material in construction. It has the characteristics of strong plasticity, good economy, high safety, and good durability. As a kind of structural material, concrete must have sufficient strength to resist various loads. At the same time, due to the brittleness of concrete, compressive strength is the most important mechanical property of concrete. To solve the disadvantages of the low efficiency of the traditional concrete compressive strength prediction methods, this study proposes a firefly algorithm (FA) and random forest (RF) hybrid machine-learning method to predict the compressive strength of concrete. First, a database is built based on the data of published articles. The dataset in the database contains eight input variables (cement, blast furnace slag, fly ash, water, superplasticizer, coarse aggregate, fine aggregate, and age) and one output variable (concrete compressive strength). Then, the correlation of the eight input variables was analyzed, and the results showed that there was no high correlation between the input variables; thus, they could be used as input variables to predict the compressive strength of concrete. Next, this study used the FA algorithm to optimize the hyperparameters of RF to obtain better hyperparameters. Finally, we verified that the FA and RF hybrid machine-learning model proposed in this study can predict the compressive strength of concrete with high accuracy by analyzing the R values and RSME values of the training set and test set and comparing the predicted value and actual value of the training set and test machine.

The new-generation information technology revolution is profoundly changing the economy and society, and promoting industrial innovation and leapfrogging. China actively promotes the synergistic development of industrial digitization and digital industrialization, which has significantly increased the scale of the digital economy. In this context, the digital transformation of enterprises has become a key issue, and the strategic choice of financial asset allocation, as an important factor affecting the operational performance, investment and financing decisions, and innovation activities of enterprises, has attracted much attention. This study constructs the framework of "financial asset allocation behavior-digital transformation", based on the data of A-share listed companies in Shanghai and Shenzhen from 2012 to 2019, and empirically analyzes it by using the fixed effect and threshold model, and finds that the relationship between the allocation of financial assets and the digital transformation is in an inverted U-shape: moderate allocation and digital transformation have an inverted U-shape relationship. It is found that the relationship between financial asset allocation and digital transformation is "inverted U-shaped": moderate allocation promotes transformation, while excessive inhibits it. Short-term financial assets play the role of "reservoir" and support the transformation, while long-term financial assets may produce crowding-out effect. Financing constraints and financial return thresholds also have differentiated impacts on the relationship. This study provides new perspectives and theoretical support for enterprises to optimize resource allocation and achieve sustainable development.

Emotions play a very vital role in human life. Emotions originate in human brain due to biological, neurological and psychological effects and influence our behavior and decision-making. This study explores the effect of environment on human emotions and behavior using questionnaire data and by applying multi-output regression. Situation based script questionnaires have been prepared to collect the responses from the participants. Questionnaires collect the demographic information, Age group, gender, and their current work in the hand at the time of responding the questions. For every question, participants have to choose one of the suitable emotions classes out of the five classes. After collecting the responses from 95 participants on five emotional classes, analysis has been conducted using three techniques i.e. descriptive analysis, single class using support vector machine and multiple classes using multi regression algorithm. Experiment results of the proposed system achieved 80% accuracy in SVM for a single class. In multi classes, results compare with different classes like emotions class with environment, emotions with age and with demographic data for multi-output regression, random forest and decision tree classifiers. Experimental results show that the proposed system performs well as compared to the entire existing system available in literature.

In recent years, cardiovascular disease (CVD) has emerged as a leading cause of mortality worldwide, highlighting the urgent need for reliable predictive models to support early diagnosis and preventive care. This study conducts a comparative analysis of various machine learning models to assess CVD risk, focusing on their ability to accurately identify individuals at high risk based on clinical and demographic data. We evaluate multiple supervised learning algorithms, including logistic regression, random forests, support vector classifiers, K-nearest neighbors, gradient boosting, and AdaBoost, comparing their predictive performance in terms of accuracy, precision, recall, and F1-score. Additionally, we propose a novel hybrid model that combines Random Forest (RF) for feature selection and Deep Neural Networks (DNN) for classification, aiming to leverage the strengths of both approaches for more accurate CVD prediction. Our findings demonstrate that ensemble models such as random forests and gradient boosting achieve superior performance, with high accuracy (0.99) and balanced precision and recall values, outperforming simpler models like logistic regression and support vector classifiers. The hybrid model further enhances prediction accuracy, achieving 92.4% accuracy, 91.7% precision, 93.0% recall, and an AUC-ROC score of 96.0%. The analysis also underscores the importance of data preprocessing techniques, including normalization and handling of missing values, in optimizing model accuracy and stability. Notably, K-nearest neighbors also performed exceptionally well with a high F1-score across classes, highlighting its robustness for this task. This study provides a detailed examination of each model's strengths and limitations, including the proposed hybrid model, offering valuable insights for healthcare practitioners and data scientists in selecting effective machine learning models for CVD risk prediction. By integrating these models into healthcare systems, real-time risk prediction can be enhanced, ultimately supporting clinical decision-making and advancing personalized care in cardiovascular health. Keywords— cardiovascular disease, machine learning, risk assessment, predictive models, supervised learning, healthcare

Anomaly detection is essential in applications such as financial fraud detection, network security, and system health monitoring. This study presents a hybrid anomaly detection model for financial transactions, combining the strengths of both unsupervised and supervised learning techniques—Isolation Forest and Random Forest, respectively. the Isolation Forest algorithm, performs unsupervised detection of outliers, effectively identifying rare or irregular data points without requiring labeled input. The detected anomalies are then used to construct a pseudo-labeled dataset, which is subsequently used to train a Random Forest classifier. The Random Forest model learns discriminative patterns between normal and anomalous data, enhancing overall detection capabilities. The system is implemented in Python using Scikit-learn for model development and evaluation. Experiments on benchmark financial transaction "Credit Card Fraud Detection Dataset from Kaggle" The hybrid model achieved a precision of 99.7%, recall of near 100%, and an F1-score of 99.98%, outperforming standalone algorithm The model also improved robustness to noise and better scalability for large datasets. Keywords: Anomaly Detection, Financial Fraud, Hybrid Machine Learning, Supervised Learning, Security, Unsupervised Learning. CISDI Journal Reference Format Azubuike Nnendah Daisy; Anireh, Vincent Ike-Emeka & Sako, D.J.S. (2025): A Model for Anomaly Detection in Financial Transactions using Hybrid Machine Learning Technique. Computing, Information Systems, Development Informatics and Allied Research Journal. Vol 16 No 3, Pp 37-50 Available online at www.isteams.net/cisdijournal. dx.doi.org/10.22624/AIMS/CISDI/V16N3P3

The rise of financial fraud, not only in India but also around the world, is a major problem. Credit card transactions have been steadily increasing in recent years, as have internet payments like Unified Payments Interface and debit card transactions, which have been increasing every day. As a result, fraud is increasing, and it has become easier for fraudsters to commit fraud. Recent research indicates the practicality of machine learning algorithms for recognizing payment fraud. Since credit cards are a common target, online fraudsters simply commit fraud because of excess use of e-commerce and other platforms, which has led to an increase in online payment methods, resulting in a greater risk of online fraud. Many researchers have started using machine learning algorithms to detect fraud. The primary objective of this study is to create a better fraud detection algorithm by analyzing payment transactional data, focusing on customers' transactional data and their payment behavioral patterns. The study proposes a framework that groups cardholders based on the volume of their transactions. Financial fraud is the biggest threat to world economies in terms of financial losses each year. The task of accurately and efficiently detecting fraud is, therefore, very challenging due to the high volume of transaction data and its complex nature. This paper presents a hybrid machine learning approach using logistic regression, decision trees, and random forests that could help enhance the accuracy and reliability of fraud detection systems. The study used an available credit card fraud dataset, with a focus on feature engineering and model evaluation to compare individual algorithms versus their ensemble. Experimental results show that the ensemble outperformed the individual classifiers in terms of precision, recall, and overall accuracy. This paper underlines the potential of hybrid machine learning techniques in improving fraud detection systems and proposes avenues for further research.