|
on Big Data |
| By: | Angel Varela |
| Abstract: | Predicting the stock market is a big challenge for the machine learning world. It is known how difficult it is to have accurate and consistent predictions with ML models. Some architectures are able to capture the movement of stocks but almost never are able to be launched to the production world. We present Achilles, with a classical architecture of LSTM(Long Short Term Memory) neural network this model is able to predict the Gold vs USD commodity. With the predictions minute-per-minute of this model we implemented a trading bot to run during 23 days of testing excluding weekends. At the end of the testing period we generated $1623.52 in profit with the methodology used. The results of our method demonstrate Machine Learning can successfully be implemented to predict the Gold vs USD commodity. |
| Date: | 2024–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2410.21291 |
| By: | Zhao, Yu |
| Abstract: | Accurately forecasting whether a real estate transaction will close is crucial for agents, lenders, and investors, impacting resource allocation, risk management, and client satisfaction. This task, however, is complex due to a combination of economic, procedural, and behavioral factors that influence transaction outcomes. Traditional machine learning approaches, particularly gradient boosting models like Gradient Boost Decision Tree, have proven effective for tabular data, outperforming deep learning models on structured datasets. However, recent advances in attention-based deep learning models present new opportunities to capture temporal dependencies and complex interactions within transaction data, potentially enhancing prediction accuracy. This article explores the challenges of forecasting real estate transaction closures, compares the performance of machine learning models, and examines how attention-based models can improve predictive insights in this critical area of real estate analytics. |
| Date: | 2024–11–08 |
| URL: | https://d.repec.org/n?u=RePEc:osf:osfxxx:sxmq2 |
| By: | Chad Brown |
| Abstract: | I consider inference in a partially linear regression model under stationary $\beta$-mixing data after first stage deep neural network (DNN) estimation. Using the DNN results of Brown (2024), I show that the estimator for the finite dimensional parameter, constructed using DNN-estimated nuisance components, achieves $\sqrt{n}$-consistency and asymptotic normality. By avoiding sample splitting, I address one of the key challenges in applying machine learning techniques to econometric models with dependent data. In a future version of this work, I plan to extend these results to obtain general conditions for semiparametric inference after DNN estimation of nuisance components, which will allow for considerations such as more efficient estimation procedures, and instrumental variable settings. |
| Date: | 2024–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2410.22574 |
| By: | Iva Glišic (National Bank of Serbia) |
| Abstract: | The paper elaborates on machine and deep learning methods, as well as mixed data sampling regression models, used for GDP nowcasting. The aim is to select an adequate model that shows better performance on the data used. The paper provides an answer to the question of whether the use of deep learning methods can improve GDP nowcasting compared to traditional econometric methods, as well as whether the use of specific high-frequency indicators improves the quality of the models used. The paper examines the selection of adequate indicators – both official and those from alternative sources, presents the framework of mixed data sampling regression models and deep learning models used for nowcasting, and gives an assessment of two such models on the example of Serbian GDP. Serbia’s GDP was modelled for the period Q1 2016 – Q2 2023 and the end of the observed period (six quarters) was used for the forecast. Finally, two assessed models were compared – the mixed data sampling regression model and the LSTM neural network. A special focus is placed on ways to improve both models. The LSTM recurrent neural network model had a smaller forecast error, with the use of a combination of official and alternative (high-frequency) indicators, but the mixed data sampling regression model also proved to be a good tool for decision-makers, since its structure allows insight into the ongoing movements impacting GDP dynamics. The use of alternative indicators in nowcasting improved the projections through both presented models. |
| Keywords: | GDP, nowcasting, MIDAS, neural networks, high-frequency indicators |
| JEL: | C32 C45 C53 |
| Date: | 2024–03 |
| URL: | https://d.repec.org/n?u=RePEc:nsb:bilten:22 |
| By: | Ali Elahi; Fatemeh Taghvaei |
| Abstract: | Predicting financial markets and stock price movements requires analyzing a company's performance, historic price movements, industry-specific events alongside the influence of human factors such as social media and press coverage. We assume that financial reports (such as income statements, balance sheets, and cash flow statements), historical price data, and recent news articles can collectively represent aforementioned factors. We combine financial data in tabular format with textual news articles and employ pre-trained Large Language Models (LLMs) to predict market movements. Recent research in LLMs has demonstrated that they are able to perform both tabular and text classification tasks, making them our primary model to classify the multi-modal data. We utilize retrieval augmentation techniques to retrieve and attach relevant chunks of news articles to financial metrics related to a company and prompt the LLMs in zero, two, and four-shot settings. Our dataset contains news articles collected from different sources, historic stock price, and financial report data for 20 companies with the highest trading volume across different industries in the stock market. We utilized recently released language models for our LLM-based classifier, including GPT- 3 and 4, and LLaMA- 2 and 3 models. We introduce an LLM-based classifier capable of performing classification tasks using combination of tabular (structured) and textual (unstructured) data. By using this model, we predicted the movement of a given stock's price in our dataset with a weighted F1-score of 58.5% and 59.1% and Matthews Correlation Coefficient of 0.175 for both 3-month and 6-month periods. |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2411.01368 |
| By: | Quechen Yang |
| Abstract: | With the increasing maturity and expansion of the cryptocurrency market, understanding and predicting its price fluctuations has become an important issue in the field of financial engineering. This article introduces an innovative Genetic Algorithm-generated Alpha Sentiment (GAS) blending ensemble model specifically designed to predict Bitcoin market trends. The model integrates advanced ensemble learning methods, feature selection algorithms, and in-depth sentiment analysis to effectively capture the complexity and variability of daily Bitcoin trading data. The GAS framework combines 34 Alpha factors with 8 news economic sentiment factors to provide deep insights into Bitcoin price fluctuations by accurately analyzing market sentiment and technical indicators. The core of this study is using a stacked model (including LightGBM, XGBoost, and Random Forest Classifier) for trend prediction which demonstrates excellent performance in traditional buy-and-hold strategies. In addition, this article also explores the effectiveness of using genetic algorithms to automate alpha factor construction as well as enhancing predictive models through sentiment analysis. Experimental results show that the GAS model performs competitively in daily Bitcoin trend prediction especially when analyzing highly volatile financial assets with rich data. |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2411.03035 |
| By: | Jesús Fernández-Villaverde (UNIVERSITY OF PENNSYLVANIA, NBER, CEPR); Galo Nuño (BANCO DE ESPAÑA, CEPR, CEMFI); Jesse Perla (UNIVERSITY OF BRITISH COLUMBIA) |
| Abstract: | We argue that deep learning provides a promising approach to addressing the curse of dimensionality in quantitative economics. We begin by exploring the unique challenges involved in solving dynamic equilibrium models, particularly the feedback loop between individual agents’ decisions and the aggregate consistency conditions required to achieve equilibrium. We then introduce deep neural networks and demonstrate their application by solving the stochastic neoclassical growth model. Next, we compare deep neural networks with traditional solution methods in quantitative economics. We conclude with a review of the applications of neural networks in quantitative economics and provide arguments for cautious optimism. |
| Keywords: | deep learning, quantitative economics |
| JEL: | C61 C63 E27 |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:bde:wpaper:2444 |
| By: | Boris Wolfson; Erman Acar |
| Abstract: | Current trends in Machine Learning prefer explainability even when it comes at the cost of performance. Therefore, explainable AI methods are particularly important in the field of Fraud Detection. This work investigates the applicability of Differentiable Inductive Logic Programming (DILP) as an explainable AI approach to Fraud Detection. Although the scalability of DILP is a well-known issue, we show that with some data curation such as cleaning and adjusting the tabular and numerical data to the expected format of background facts statements, it becomes much more applicable. While in processing it does not provide any significant advantage on rather more traditional methods such as Decision Trees, or more recent ones like Deep Symbolic Classification, it still gives comparable results. We showcase its limitations and points to improve, as well as potential use cases where it can be much more useful compared to traditional methods, such as recursive rule learning. |
| Date: | 2024–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2410.21928 |
| By: | Chris Lam |
| Abstract: | Alternative data provides valuable insights for lenders to evaluate a borrower's creditworthiness, which could help expand credit access to underserved groups and lower costs for borrowers. But some forms of alternative data have historically been excluded from credit underwriting because it could act as an illegal proxy for a protected class like race or gender, causing redlining. We propose a method for applying causal inference to a supervised machine learning model to debias alternative data so that it might be used for credit underwriting. We demonstrate how our algorithm can be used against a public credit dataset to improve model accuracy across different racial groups, while providing theoretically robust nondiscrimination guarantees. |
| Date: | 2024–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2410.22382 |
| By: | Matteo Citterio; Marco D'Errico; Gabriele Visentin |
| Abstract: | We introduce a novel Dynamic Graph Neural Network (DGNN) architecture for solving conditional $m$-steps ahead forecasting problems in temporal financial networks. The proposed DGNN is validated on simulated data from a temporal financial network model capturing stylized features of Interest Rate Swaps (IRSs) transaction networks, where financial entities trade swap contracts dynamically and the network topology evolves conditionally on a reference rate. The proposed model is able to produce accurate conditional forecasts of net variation margins up to a $21$-day horizon by leveraging conditional information under pre-determined stress test scenarios. Our work shows that the network dynamics can be successfully incorporated into stress-testing practices, thus providing regulators and policymakers with a crucial tool for systemic risk monitoring. |
| Date: | 2024–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2410.23275 |
| By: | Mahdi Goldani |
| Abstract: | This study focuses on predicting the Human Development Index (HDI) trends for GCC countries Saudi Arabia, Qatar, Kuwait, Bahrain, United Arab Emirates, and Omanusing machine learning techniques, specifically the XGBoost algorithm. HDI is a composite measure of life expectancy, education, and income, reflecting overall human development. Data was gathered from official government sources and international databases, including the World Bank and UNDP, covering the period from 1996 to 2022. Using the Edit Distance on Real sequence (EDR) method for feature selection, the model analyzed key indicators to predict HDI values over the next five years (2023-2027). The model demonstrated strong predictive accuracy for in-sample data, but minor overfitting issues were observed with out-of-sample predictions, particularly in the case of the UAE. The forecast results suggest that Kuwait, Bahrain, and the UAE will see stable or slightly increasing HDI values, while Saudi Arabia, Qatar, and Oman are likely to experience minimal fluctuations or slight decreases. This study highlights the importance of economic, health, and educational indicators in determining HDI trends and emphasizes the need for region-specific predictive models to improve accuracy. Policymakers should focus on targeted interventions in healthcare, education, and economic diversification to enhance human development outcomes. |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2411.01177 |
| By: | Du, Tianyu (Stanford U); Kanodia, Ayush (Stanford U); Brunborg, Herman (Stanford U); Vafa, Keyon (Harvard U); Athey, Susan (Stanford U) |
| Abstract: | Many empirical studies of labor market questions rely on estimating relatively simple predictive models using small, carefully constructed longitudinal survey datasets based on hand-engineered features. Large Language Models (LLMs), trained on massive datasets, encode vast quantities of world knowledge and can be used for the next job prediction problem. However, while an off-the-shelf LLM produces plausible career trajectories when prompted, the probability with which an LLM predicts a particular job transition conditional on career history will not, in general, align with the true conditional probability in a given population. Recently, Vafa et al. (2024) introduced a transformer-based “foundation model†, CAREER, trained using a large, unrepresentative resume dataset, that predicts transitions between jobs; it further demonstrated how transfer learning techniques can be used to leverage the foundation model to build better predictive models of both transitions and wages that reflect conditional transition probabilities found in nationally representative survey datasets. This paper considers an alternative where the fine-tuning of the CAREER foundation model is replaced by fine-tuning LLMs. For the task of next job prediction, we demonstrate that models trained with our approach outperform several alternatives in terms of predictive performance on the survey data, including traditional econometric models, CAREER, and LLMs with in-context learning, even though the LLM can in principle predict job titles that are not allowed in the survey data. Further, we show that our fine-tuned LLM-based models’ predictions are more representative of the career trajectories of various workforce subpopulations than off-the-shelf LLM models and CAREER. We conduct experiments and analyses that highlight the sources of the gains in the performance of our models for representative predictions. |
| Date: | 2024–06 |
| URL: | https://d.repec.org/n?u=RePEc:ecl:stabus:4188 |
| By: | Yu, Shubin |
| Abstract: | This study explores the efficacy of large language models (LLMs) in short-text topic modeling, comparing their performance with human evaluation and Latent Dirichlet Allocation (LDA). In Study 1, we analyzed a dataset on chatbot anthropomorphism using human evaluation, LDA, and two LLMs (GPT-4 and Claude). Results showed that LLMs produced topic classifications similar to human analysis, outperforming LDA for short texts. In Study 2, we investigated the impact of sample size and LLM choice on topic modeling consistency using a COVID-19 vaccine hesitancy dataset. Findings revealed high consistency (80-90%) across various sample sizes, with even a 5% sample achieving 90% consistency. Comparison of three LLMs (Gemini Pro 1.5, GPT-4o, and Claude 3.5 Sonnet) showed comparable performance, with two models achieving 90% consistency. This research demonstrates that LLMs can effectively perform short-text topic modeling in medical informatics, offering a promising alternative to traditional methods. The high consistency with small sample sizes suggests potential for improved efficiency in research. However, variations in performance highlight the importance of model selection and the need for human supervision in topic modeling tasks. |
| Date: | 2024–11–01 |
| URL: | https://d.repec.org/n?u=RePEc:osf:osfxxx:mqk3r |
| By: | Kentaro Hoshisashi; Carolyn E. Phelan; Paolo Barucca |
| Abstract: | Calibrating the time-dependent Implied Volatility Surface (IVS) using sparse market data is an essential challenge in computational finance, particularly for real-time applications. This task requires not only fitting market data but also satisfying a specified partial differential equation (PDE) and no-arbitrage conditions modelled by differential inequalities. This paper proposes a novel Physics-Informed Neural Networks (PINNs) approach called Whack-a-mole Online Learning (WamOL) to address this multi-objective optimisation problem. WamOL integrates self-adaptive and auto-balancing processes for each loss term, efficiently reweighting objective functions to ensure smooth surface fitting while adhering to PDE and no-arbitrage constraints and updating for intraday predictions. In our experiments, WamOL demonstrates superior performance in calibrating intraday IVS from uneven and sparse market data, effectively capturing the dynamic evolution of option prices and associated risk profiles. This approach offers an efficient solution for intraday IVS calibration, extending PINNs applications and providing a method for real-time financial modelling. |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2411.02375 |
| By: | Shapeng Jiang; Lijia Wei; Chen Zhang |
| Abstract: | In recent years, large language models (LLMs) have attracted attention due to their ability to generate human-like text. As surveys and opinion polls remain key tools for gauging public attitudes, there is increasing interest in assessing whether LLMs can accurately replicate human responses. This study examines the potential of LLMs, specifically ChatGPT-4o, to replicate human responses in large-scale surveys and to predict election outcomes based on demographic data. Employing data from the World Values Survey (WVS) and the American National Election Studies (ANES), we assess the LLM's performance in two key tasks: simulating human responses and forecasting U.S. election results. In simulations, the LLM was tasked with generating synthetic responses for various socio-cultural and trust-related questions, demonstrating notable alignment with human response patterns across U.S.-China samples, though with some limitations on value-sensitive topics. In prediction tasks, the LLM was used to simulate voting behavior in past U.S. elections and predict the 2024 election outcome. Our findings show that the LLM replicates cultural differences effectively, exhibits in-sample predictive validity, and provides plausible out-of-sample forecasts, suggesting potential as a cost-effective supplement for survey-based research. |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2411.01582 |
| By: | Hardik Routray; Bernhard Hientzsch |
| Abstract: | We propose a simple methodology to approximate functions with given asymptotic behavior by specifically constructed terms and an unconstrained deep neural network (DNN). The methodology we describe extends to various asymptotic behaviors and multiple dimensions and is easy to implement. In this work we demonstrate it for linear asymptotic behavior in one-dimensional examples. We apply it to function approximation and regression problems where we measure approximation of only function values (``Vanilla Machine Learning''-VML) or also approximation of function and derivative values (``Differential Machine Learning''-DML) on several examples. We see that enforcing given asymptotic behavior leads to better approximation and faster convergence. |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2411.05257 |
| By: | Joanna AL JOCMEK (GRANEM - Groupe de Recherche Angevin en Economie et Management - UA - Université d'Angers - Institut Agro Rennes Angers - Institut Agro - Institut national d'enseignement supérieur pour l'agriculture, l'alimentation et l'environnement, Centre Hospitalier du Nord, Zgharta, Lebanon); Mazen MOUSSALLEM (Holy Family University, Batroun, Lebanon) |
| Abstract: | Managerial Innovation (MI), which involves adopting new management practices or methods essential for enhancing organizational performance, faces limitations, particularly in reducing the number of patients who do not pay their bills. This paper aims to address this issue by developing an Artificial Intelligence (AI) model that predicts early which patients are likely to default on their payments. By doing so, stronger MI strategies can be applied specifically to these patients, thereby avoiding the adverse impact on those expected to fulfill their financial obligations.In this study, data from two patient groups were used: 683 patients from internal admissions and 753 patients from urgent admissions. The "KNeighborsClassifier" machine learning algorithm was trained and tested 10, 000 times on each dataset, each time with 90% of the data randomly selected for training and the remaining 10% for evaluation.By leveraging these AI predictions, it is possible to implement more focused MI procedures: applied to only 18 out of 100 internal patients, reducing the number of unpaid patients from 10 to 5; and applied to only 10 out of 100 urgent patients, reducing the number of unpaid patients from 6 to 5. Using a larger patient dataset could further enhance these results. |
| Abstract: | L'Innovation Managériale (IM), qui consiste à adopter de nouvelles pratiques ou méthodes de gestion essentielles pour améliorer la performance organisationnelle, présente des limitations, notamment en ce qui concerne la réduction du nombre de patients qui ne paient pas leurs factures. Cet article vise à résoudre ce problème en développant un modèle d'Intelligence Artificielle (IA) capable de prédire de manière précoce quels patients sont susceptibles de ne pas régler leurs paiements. Ce faisant, des stratégies d'IM plus ciblées peuvent être appliquées spécifiquement à ces patients, évitant ainsi les impacts négatifs sur ceux qui sont censés remplir leurs obligations financières. Dans cette étude, les données de deux groupes de patients ont été utilisées : 683 patients provenant des admissions internes et 753 patients provenant des admissions urgentes. L'algorithme d'apprentissage automatique "KNeighborsClassifier" a été entraîné et testé 10 000 fois sur chaque ensemble de données, à chaque fois avec 90 % des données sélectionnées aléatoirement pour l'entraînement et les 10 % restants pour l'évaluation. En utilisant ces prédictions d'IA, il est possible de mettre en oeuvre des procédures d'IM plus ciblées : appliquées seulement à 18 patients sur 100 dans les admissions internes, réduisant le nombre de patients non réglés de 10 à 5 ; et appliquées seulement à 10 patients sur 100 dans les admissions urgentes, réduisant le nombre de patients non réglés de 6 à 5. L'utilisation d'un ensemble de données plus large pourrait améliorer davantage ces résultats. |
| Keywords: | Artificial Intelligence -Managerial Innovation -Machine Learning -Payment Prediction -Unpaid Patient Bills. Alternatives Managériales, Intelligence Artificielle - Innovation Managériale - Apprentissage Automatique - Prédiction des Paiements - Factures Impayées des Patients. |
| Date: | 2024–11–05 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-04770751 |
| By: | Mirko Djukic (National Bank of Serbia) |
| Abstract: | The frequency of certain topics in newspaper articles can be a good indicator of some economic developments. The application of topic modelling in the Serbian language, using the LDA model, is hampered by the fact that Serbian is a highly inflectional language, where words have a large number of forms which the model recognises as words with a different meaning. In this paper, we tried to turn that aggravating circumstance into an advantage by reducing only the economic words to their base form. Thus, we attributed to them a greater relevance than to non-economic words, which remained in a large number of forms with a lower frequency of occurrence. As the topics classified in this manner were mostly based on economic expressions, it was expected that they would have a greater applicability in further economic analyses. |
| Keywords: | textual analysis, topic modelling, Latent Dirichlet Allocation, LASSO model |
| JEL: | C13 C55 E31 E37 E52 |
| Date: | 2024–03 |
| URL: | https://d.repec.org/n?u=RePEc:nsb:bilten:21 |