Volume 193
Published on July 2025Volume title: Proceedings of ICEMGD 2025 Symposium: Innovating in Management and Economic Development

It is difficult to predict stock prices due to volatile financial markets and various economic factors. However, useful effective predictive techniques can offer great assistance to analysts and investors, and therefore much research keeps being conducted in this area. This study analyzed the predictive capabilities of two popular deep learning methods: Long Short-Term Memory (LSTM) and Multi-Layer Perceptron (MLP), based on a historical stock price dataset. This study conducted experiments through the daily stock prices of Google and a larger dataset that covered multiple international companies. It also evaluated the rigorousness of LSTM and MLP under many conditions, such as different fluctuation mechanisms, high and low price levels, and datasets of varying scales. The study found that the prediction accuracy of both LSTM and MLP was satisfactory. However, during stable and low-fluctuation periods, LSTM achieved better performance than MLP, but on smaller datasets, MLP showed stronger generalization capabilities. Therefore, to improve predictive capabilities, which model to use should be based on market context and data scale.

Under the dual challenges of fintech evolution and digital transformation, commercial banks face increasing limitations in traditional marketing prediction methods, which struggle with static customer profiling, low data utilization, and poor adaptability to real-time demands. This study addresses these gaps by proposing an XGBoost-based predictive framework to enhance precision marketing and risk-adjusted returns in banking scenarios. We integrate multidimensional features, including static attributes (e.g., age, occupation) and dynamic behavioral indicators (e.g., consumer confidence index, Euribor rates), to overcome the unidimensional profiling limitations of conventional approaches. Methodologically, XGBoost demonstrates superior performance through three innovations: firstly efficient handling of high-dimensional sparse data via parallel computing, reducing marginal processing costs while improving prediction accuracy (89% accuracy, 90% AUC). Secondly, mitigation of information asymmetry by synthesizing transactional, social, and macroeconomic features (e.g., employment variation rate, housing loans). Comparative analyses against five benchmark models (GBDT, Random Forest, Decision Tree, Logistic Regression, Bagging) confirm XGBoost’s dominance in AUC and F1-score, validating its capacity to resolve nonlinear interactions and temporal sensitivity in marketing campaigns. The model’s scalability enables cost-effective targeting. This research contributes to both algorithmic optimization in financial marketing and operational decision-making frameworks, though limitations persist in handling extreme class imbalances. Future work will explore hybrid architectures combining XGBoost with deep learning for cross-channel behavioral modeling.
Climate change has been one of the most pressing issues of this century, and it requires urgent and concerted global action. In the meantime, technology has become a critical tool in mitigating and adapting to its impacts. This essay, through a method of literature review, explores how various technological innovations, such as renewable energy, carbon capture, artificial intelligence, and smart infrastructure, play a role in mitigating greenhouse gas emissions, fostering climate resilience, and promoting energy transitions. Through case studies and secondary research, this paper demonstrates how a wide range of technological innovations help reduce carbon emissions and strengthen climate resilience. The findings underlined the importance of technological innovations in transitioning to a low-carbon economy while also revealing some existing barriers such as high cost, policy gaps, and social resistance. Finally, the essay argues that global cooperation, a strong policy framework, and equitable use of technological solutions are essential for further success.

With the rapid development of consumer finance, personal credit business is becoming increasingly active in the financial market, but it is also accompanied by significant credit risks.How to use big data methods to achieve efficient and accurate default prediction has become a core issue in the field of credit risk control.This article takes the Home Credit Default Risk dataset on the Kaggle platform as the research object, and uses data mining methods to systematically analyze the statistical correlation between external credit scores, borrower annual income, loan amounts, and other key features and default risk.By using Information Value (IV) and Kernel Density Estimation (KDE) methods to screen high discriminative force variables, and based on data distribution characteristics, a prediction model with Random Forest and Extreme Gradient Boosting Tree (XGBoost) as the core is constructed to compare its performance under precision, recall, F1 value, and AUC indicators.The results show that XGBoost has better recognition ability in imbalanced data scenarios, while random forests have more advantages in feature interpretability.The research results not only verify the effectiveness of the feature distribution driven model design, but also provide practical suggestions and theoretical support for financial institutions in pre loan risk screening and dynamic monitoring during loans.
This paper compares the financial performance, market strategies and positioning of the four key players in the competitive athletic apparel industry; Nike, Lululemon, Under Armour, and Athleta, The Gap. This study evaluates how each business creates and preserves its competitive advantage in the market where rapid changing consumer demand occurs through SWOT analysis. The findings reveal that by integrating wellness and strong community ties to build loyalty and maintain growth. Nike continues to dominate the world market through its excellent marketing strategies of emotional marketing, female-centric marketing and sponsorships that resonate globally. Although Under Armour struggles with consistent branding, its tech-driven designs appeal to young people. As a subsidiary of Gap Inc., Athleta advocates for eco-consciousness and women empowered messaging, but its brand recognition and unique selling points are still quite low compared to its larger competitors. This study helps to bridge the gap between consumer behaviour, gender-focused branding, and financial analysis. The innovation of the paper to evaluate brand competitiveness comprehensively, is to combine financial health comparison with strategic marketing evaluation.
Nansha Economic and Technological Development Zone, as a key strategic platform within the Guangdong-Hong Kong-Macau Greater Bay Area (GBA), serves as a vital engine for regional economic growth and industrial upgrading. This paper delves into the integration pathway of “port logistics—port-proximate industry—technological innovation” in Nansha and the pivotal role of policy instruments in spurring regional economic transformation and upgrading. It unveils the rationale underpinning Nansha’s evolution from a conventional industrial base to a hub for new productive forces. The analysis reveals that Nansha must refine policy execution, bolster cross-regional collaboration, and enhance industrial chain coordination. The paper provides recommendations such as streamlining policy procedures and upgrading education and healthcare infrastructure, offering insights into institutional innovation and industrial integration within regional development zones. Furthermore, Nansha’s strategic significance lies not only in its role as a driver for regional economic growth, but also in its potential to serve as a model for innovation and cooperation in the global economic landscape. Additionally, it underscores the significance of Nansha’s strategic advantages in the context of China’s “dual circulation” development strategy, providing a reference for other regional development zones striving for high-quality development through policy and industrial integration.

This study focuses on the prediction of the Environmental Kuznets Curve (EKC) and inflection point issues in the Yangtze River Pearl River Delta basin in the future, and conducts an in-depth analysis by comprehensively applying multiple research methods. By combing the theoretical evolution of the Environmental Kuznets Curve, this study clarifies its historical process from proposal to development and its morphological changes in different contexts. Based on the current status of China's economy and environment, it is found that there is a complex relationship between economic growth and environmental pollution at the current stage. Although the emissions of some pollutants show a downward trend, the overall situation remains severe, and the timing of the inflection point is constrained by multiple factors. On the basis of analyzing influencing factors such as industrial structure, technological innovation, and environmental policies, this study constructs a dynamic computable general equilibrium model (CGE) to simulate the impacts of different policy combinations on the curve's inflection point, puts forward targeted policy recommendations to promote the coordinated development of economy and environment, and provides theoretical support and decision-making basis for China to achieve sustainable development.

The financial market is always changing rapidly, and it is crucial for people to monitor certain activities to ensure the overall safety of transactions or exchanges. The applications of machine learning have delved into many areas include but not limited to risk management, natural language processing, computer vision etc. This paper explores the application of five classic Machine Learning (ML) algorithms, including Decision Tree, Random Forest, Logistic Regression, XGBoost and Support Vector Machine (SVM) on loan application to test their raw performances on an imbalanced dataset, and comparing their results to determine which models behaves most ideally under very little optimization made. The project reveals common weaknesses for machine learning models when applying on some imbalanced dataset, where the recall and f1-score are not as ideal as accuracy and precision as they are biased toward the majority samples. However, models such as Random Forest does produce a desirable result, thus offering valuable reference under the context of determining the eligibility of loan applicants.

Stock market investment involves various levels of risk. This study attempts to solve the problem of classifying stocks into specific risk categories using fundamental and technical indicators. This study helps bridge the gap between theoretical classification of risks and practical trading approaches for various investment styles. The author employs quantitative methods such as descriptive statistics, risk scoring, time series analysis, and Auto-Regressive Integrated Moving Average (ARIMA) modeling on 10 diverse U.S. stocks from January 2020 to current date. These analyses provided evidence of three risk categories with distinct volatility patterns. Relative Strength Index (RSI) has been found to be the most statistically relevant variable for all returns, while volatility had the largest absolute coefficient of returns. The author found that ARIMA(0,1,0) best fit all stocks. This indicated that all stocks followed a random walk with drift, regardless of risk category, but with regular risk stocks exhibiting slightly more predictability than conservative or aggressive. These results will be useful to portfolio managers and investors who wish to adjust their stock portfolios according to their risk appetite by helping align stock selection with risk tolerance while indicating which technical indicators are useful for different styles.
The Guangdong-Hong Kong-Macau Greater Bay Area (GBA) achieves the goal of integration and sustainable development of all regions through the creation of comprehensive smart cities. Leveraging the smart city effect plays a crucial role in the development of the GBA. However, the key to creating a good smart city in each region lies in whether the region’s public finance system can be effectively utilised to achieve the fundamental objective of promoting smart city development through the key role of digital technology for governance, infrastructure and innovation. This study examines findings on local development disparities in the GBA and their impact on infrastructure, public services, and industrial innovation, and proposes measures to improve cross-regional fiscal coordination. The measures are developed through case studies and policy comparisons, with a focus on Shenzhen’s intelligent transport. Recommendations include the establishment of a joint municipal financing mechanism, the promotion of inter-city cooperation, and the enhancement of private sector participation through a public-private partnership model. The paper highlights the importance of establishing institutionalised fiscal relations to ensure the building of equitable and efficient smart cities in the GBA.