nep-cmp New Economics Papers
on Computational Economics
Issue of 2023‒01‒02
ten papers chosen by

  1. pystacked: Stacking generalization and machine learning in Stata By Christian B. Hansen; Mark E. Schaffer; Achim Ahrens
  2. Asymptotic study of stochastic adaptive algorithm in non-convex landscape By Sébastien Gadat; Ioana Gavra
  4. Data-gravity and Data Science: Educational Approaches and Solutions By Popov Alexandr; Deryabin Andrey; Gluhov Pavel
  5. Using Machine Learning for Efficient Flexible Regression Adjustment in Economic Experiments By John List; Ian Muir; Gregory Sun
  6. A probability transducer and decision-theoretic augmentation for machine-learning classifiers By Dyrland, Kjetil; Lundervold, Alexander Selvikvåg; Porta Mana, PierGianLuca
  8. The Short-Term Predictability of Returns in Order Book Markets: a Deep Learning Perspective By Lorenzo Lucchese; Mikko Pakkanen; Almut Veraart
  9. Identifying and characterising AI adopters: A novel approach based on big data By Flavio Calvino; Lea Samek; Mariagrazia Squicciarini; Cody Morris
  10. Mod-Poisson approximation schemes: Applications to credit risk By Pierre-Lo\"ic M\'eliot; Ashkan Nikeghbali; Gabriele Visentin

  1. By: Christian B. Hansen (University of Chicago); Mark E. Schaffer (Heriot-Watt University); Achim Ahrens (ETH Zürich)
    Abstract: pystacked implements stacked generalization (Wolpert 1992) for regression and binary classification via Python’s scikit-learn. Stacking combines multiple supervised machine learners—the “base” or “level-0” learners—into a single learner. The currently supported base learners include regularized regression, random forest, gradient boosted trees, support vector machines, and feed-forward neural nets (multilayer perceptron). pystacked can also be used as a ‘regular’ machine learning program to fit a single base learner and, thus, provides an easy-to-use API for scikit-learn’s machine learning algorithms.
    Date: 2022–11–30
  2. By: Sébastien Gadat (TSE-R - Toulouse School of Economics - UT1 - Université Toulouse 1 Capitole - Université Fédérale Toulouse Midi-Pyrénées - EHESS - École des hautes études en sciences sociales - CNRS - Centre National de la Recherche Scientifique - INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement); Ioana Gavra (IRMAR - Institut de Recherche Mathématique de Rennes - UR1 - Université de Rennes 1 - UNIV-RENNES - Université de Rennes - INSA Rennes - Institut National des Sciences Appliquées - Rennes - INSA - Institut National des Sciences Appliquées - UNIV-RENNES - Université de Rennes - ENS Rennes - École normale supérieure - Rennes - UR2 - Université de Rennes 2 - UNIV-RENNES - Université de Rennes - CNRS - Centre National de la Recherche Scientifique - Institut Agro Rennes Angers - Institut Agro - Institut national d'enseignement supérieur pour l'agriculture, l'alimentation et l'environnement)
    Abstract: This paper studies some asymptotic properties of adaptive algorithms widely used in optimization and machine learning, and among them Adagrad and Rmsprop, which are involved in most of the blackbox deep learning algorithms. Our setup is the non-convex landscape optimization point of view, we consider a one time scale parametrization and we consider the situation where these algorithms may be used or not with mini-batches. We adopt the point of view of stochastic algorithms and establish the almost sure convergence of these methods when using a decreasing step-size towards the set of critical points of the target function. With a mild extra assumption on the noise, we also obtain the convergence towards the set of minimizers of the function. Along our study, we also obtain a \convergence rate" of the methods, in the vein of the works of [GL13].
    Keywords: Stochastic optimization,Stochastic adaptive algorithm,Convergence of random variables
    Date: 2022–08
  3. By: Le, Anh Hoang (Ho Chi Minh University of Banking); , Le Nguyen Hoai Thi; Huong, Luong Tran Hoang; , La Phu Hao; Nga, Nguyen Thi Thuy
    Abstract: The client experience has been improved by the recent growth of digital banking services (PwC, 2018). Finding the variables that influence how customers experience this service is the issue that now interests researchers and commercial banks. This study focuses on identifying the factors impacting consumers' experiences with digital banking services at Vietnamese commercial banks in an effort to provide a solution to the aforementioned problem. This study is also the first to combine interaction estimation through a structural equation modeling (SEM), and machine learning techniques through an artificial neural network (ANN) model to create a predictive model of customer experience on digital banking services in Vietnamese commercial banks. The SEM model estimation results indicate that perceived convenience, functional quality, and service quality, brand awareness, safety perception, and usability are the elements influencing the customer's experience utilizing digital banking services. In order to improve the customer experience of digital banking services at Vietnamese commercial banks, the study has developed a customer experience forecasting model and provided some managerial implications.
    Date: 2022–11–13
  4. By: Popov Alexandr (Russian Presidential Academy of National Economy and Public Administration); Deryabin Andrey (Russian Presidential Academy of National Economy and Public Administration); Gluhov Pavel (Russian Presidential Academy of National Economy and Public Administration)
    Abstract: In the context of fundamental changes in the economy and the labor market, the introduction of educational programs in the field of data analysis and machine learning at all levels of education with the priority of integrating mathematical, natural science and socio-humanitarian knowledge becomes important. An overview and analysis of foreign experience and main discussion topics in the development of educational modules for data science and machine learning for adolescents and adolescents is presented.
    Keywords: data science; machine learning; educational programs; data literacy; data analysis; data science; artificial Intelligence; general education; vocational guidance; additional education; computer science; STEM; social science
    Date: 2021–01
  5. By: John List; Ian Muir; Gregory Sun
    Abstract: This study investigates how to use regression adjustment to reduce variance in experimental data. We show that the estimators recommended in the literature satisfy an orthogonality property with respect to the parameters of the adjustment. This observation greatly simplifies the derivation of the asymptotic variance of these estimators and allows us to solve for the efficient regression adjustment in a large class of adjustments. Our efficiency results generalize a number of previous results known in the literature. We then discuss how this efficient regression adjustment can be feasibly implemented. We show the practical relevance of our theory in two ways. First, we use our efficiency results to improve common practices currently employed in field experiments. Second, we show how our theory allows researchers to robustly incorporate machine learning techniques into their experimental estimators to minimize variance.
    Date: 2022
  6. By: Dyrland, Kjetil; Lundervold, Alexander Selvikvåg (Western Norway University of Applied Sciences); Porta Mana, PierGianLuca (HVL Western Norway University of Applied Sciences)
    Abstract: In a classification task from a set of features, one would ideally like to have the probability of the class conditional on the features. Such probability is computationally almost impossible to find in many important cases. The primary idea of the present work is to calculate the probability of a class conditional not on the features, but on a trained classifying algorithm's output. Such probability is easily calculated and provides an output-to-probability ’transducer’ that can be applied to the algorithm's future outputs. In conjunction with problem-dependent utilities, the probabilities of the transducer allows one to make the optimal choice among the classes or among a set of more general decisions, by means of expected-utility maximization. The combined procedure is a computationally cheap yet powerful ‘augmentation’ of the original classifier. This idea is demonstrated in a simplified drug-discovery problem with a highly imbalanced dataset. The augmentation leads to improved results, sometimes close to theoretical maximum, for any set of problem-dependent utilities. The calculation of the transducer also provides, automatically: (i) a quantification of the uncertainty about the transducer itself; (ii) the expected utility of the augmented algorithm (including its uncertainty), which can be used for algorithm selection; (iii) the possibility of using the algorithm in a ‘generative mode’, useful if the training dataset is biased. It is argued that the optimality, flexibility, and uncertainty assessment provided by the transducer & augmentation are dearly needed for classification problems in fields such as medicine and drug discovery.
    Date: 2022–06–01
  7. By: Olayiwola Oladiran; Muhammad Abbas
    Abstract: This paper examines the relationship between student housing attributes and the pricing of student accommodation. The paper further explores the asymmetries in pricing for Purpose- built Student Accommodation (PBSA) and Private Student Accommodation Providers (PSAP). We utilise a web scraping procedure to access online-listed property information and prices from 25 major student destination cities in Europe on and Study Abroad Apartments. Using machine learning methodology, we analyse some key tangible and non-tangible features of the properties and explore their relationships with the listed price. We also examine the potential effects of economies of scale through variations in the pricing mechanism for PBSAs and PSAPs. The results show that the non-tangible property attributes have a stronger relationship with student accommodation prices in comparison to the tangible attributes. We also observe that the influence of these non-tangible property features on student accommodation prices is significantly stronger for PSAP properties in comparison to PBSA properties. The results suggest that through the economies of scale mechanism, institutional investors may be able to provide some facilities in their PBSAs at lower costs than PSAP investors and this may result in lower premiums for these facilities as reflected in the pricing. From a methodological point of view, we show that the use of asset features and historic pricing trends can enable the training of various supervised machine learning algorithms which in turn can improve asset pricing, taking account of national and non-institutional investment types.
    JEL: R3
    Date: 2022–01–01
  8. By: Lorenzo Lucchese; Mikko Pakkanen; Almut Veraart
    Abstract: In this paper, we conduct a systematic large-scale analysis of order book-driven predictability in high-frequency returns by leveraging deep learning techniques. First, we introduce a new and robust representation of the order book, the volume representation. Next, we conduct an extensive empirical experiment to address various questions regarding predictability. We investigate if and how far ahead there is predictability, the importance of a robust data representation, the advantages of multi-horizon modeling, and the presence of universal trading patterns. We use model confidence sets, which provide a formalized statistical inference framework particularly well suited to answer these questions. Our findings show that at high frequencies predictability in mid-price returns is not just present, but ubiquitous. The performance of the deep learning models is strongly dependent on the choice of order book representation, and in this respect, the volume representation appears to have multiple practical advantages.
    Date: 2022–11
  9. By: Flavio Calvino; Lea Samek; Mariagrazia Squicciarini; Cody Morris
    Abstract: This work employs a novel approach to identify and characterise firms adopting Artificial Intelligence (AI), using different sources of large microdata. Focusing on the United Kingdom, the analysis combines data on Intellectual Property Rights, website information, online job postings, and firm-level financials for the first time. It shows that a significant share of AI adopters is active in Information and Communication Technologies and professional services, and is located in the South of the United Kingdom, particularly around London. Adopters tend to be highly productive and larger than other firms, while young adopters tend to hire AI workers more intensively. Human capital appears to play an important role, not only for AI adoption but also for firms’ productivity returns. Significant differences in the characteristics of AI adopters emerge when distinguishing between firms carrying out AI innovation, those with an AI core business, and those searching for AI talent.
    Keywords: artificial intelligence, productivity, technology adoption
    Date: 2022–12–19
  10. By: Pierre-Lo\"ic M\'eliot; Ashkan Nikeghbali; Gabriele Visentin
    Abstract: We introduce a new numerical approximation method for functionals of factor credit portfolio models based on the theory of mod-$\phi$ convergence and mod-$\phi$ approximation schemes. The method can be understood as providing correction terms to the classic Poisson approximation, where higher order corrections lead to asymptotically better approximations as the number of obligors increases. We test the model empirically on two tasks: the estimation of risk measures ($\mathrm{VaR}$ and $\mathrm{ES}$) and the computation of CDO tranche prices. We compare it to other commonly used methods -- such as the recursive method, the large deviations approximation, the Chen--Stein method and the Monte Carlo simulation technique (with and without importance sampling) -- and we show that it leads to more accurate estimates while requiring less computational time.
    Date: 2022–10

General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.