A fast and scalable ensemble of global models with long memory and data partitioning for the M5 forecasting competition

This work presents key insights on the model development strategies used in our cross-learning-based retail demand forecast framework. The proposed framework outperforms state-of-the-art univariate models in the time series forecasting literature. It has achieved 17th position in the accuracy track of the M5 forecasting competition, which is among the top 1% of solutions.     

Diagnosing a healthcare cybersecurity crisis: The impact of IoMT advancements and 5G

Internet of Medical Things (IoMT) technology remains in early stages of adoption, but advancements and breakthroughs are quickly moving this process forward. There is a critical need for cybersecurity to be a priority in the development of these new tools, alongside design and utility. Given the rapid pace and potential magnitude of the coming advancements in IoMT, if privacy and security risks are neglected, a significant crisis could emerge in the form of more frequent cybersecurity breaches. This article examines the market opportunities and risks associated with IoMT and outlines a plan for proactively mitigating concerns and providing a platform to foster growth, to modify attitudes and behaviors, and to continue to build consumer confidence in the overall health system without sacrificing security.     

M5 competition uncertainty: Overdispersion,distributional forecasting,GAMLSS, and beyond

The M5 competition uncertainty track aims for probabilistic forecasting of sales of thousands of Walmart retail goods. We show that the M5 competition data face strong overdispersion and sporadic demand, especially zero demand. We discuss modeling issues concerning adequate probabilistic forecasting of such count data processes. Unfortunately, the majority of popular prediction methods used in the M5 competition (e.g. lightgbm and xgboost GBMs) fail to address the data characteristics, due to the considered objective functions. Distributional forecasting provides a suitable modeling approach to overcome those problems. The GAMLSS framework allows for flexible probabilistic forecasting using low-dimensional distributions. We illustrate how the GAMLSS approach can be applied to M5 competition data by modeling the location and scale parameters of various distributions, e.g. the negative binomial distribution. Finally, we discuss software packages for distributional modeling and their drawbacks, like the R package gamlss with its package extensions, and (deep) distributional forecasting libraries such as TensorFlow Probability.     

Understanding machine learning-based forecasting methods: A decomposition framework and research opportunities

Machine learning (ML) methods are gaining popularity in the forecasting field, as they have shown strong empirical performance in the recent M4 and M5 competitions, as well as in several Kaggle competitions. However, understanding why and how these methods work well for forecasting is still at a very early stage, partly due to their complexity. In this paper, I present a framework for regression-based ML that provides researchers with a common language and abstraction to aid in their study. To demonstrate the utility of the framework, I show how it can be used to map and compare ML methods used in the M5 Uncertainty competition. I then describe how the framework can be used together with ablation testing to systematically study their performance. Lastly, I use the framework to provide an overview of the solution space in regression-based ML forecasting, identifying areas for further research.     

Exploring the representativeness of the M5 competition data

The main objective of the M5 competition, which focused on forecasting the hierarchical unit sales of Walmart, was to evaluate the accuracy and uncertainty of forecasting methods in the field to identify best practices and highlight their practical implications. However, can the findings of the M5 competition be generalized and exploited by retail firms to better support their decisions and operation? This depends on the extent to which M5 data is sufficiently similar to unit sales data of retailers operating in different regions selling different product types and considering different marketing strategies. To answer this question, we analyze the characteristics of the M5 time series and compare them with those of two grocery retailers, namely Corporación Favorita and a major Greek supermarket chain, using feature spaces. Our results suggest only minor discrepancies between the examined data sets, supporting the representativeness of the M5 data.     

Understanding expert practices in order to control expert activities: The case of trading

Experts are important actors of organizational control. Nevertheless, experience suggests that they must be controlled as well. This is particularly the case for traders in financial institutions. We first identify the limits of traditional control patterns when the managing the activities of experts is at stake. Hyperspecialization, which is the ability to act within different logics and multiple time horizons, suggests that multidimensional representations of these activities be adopted and made explicit, which has the potential to prevent such activities from turning problematic. By examining bank risks and conducting additional interviews with actors from bank trading services, we recommend that multiple components of complexity be preserved when dealing with expert‐related operational risks, instead of reducing this complexity to a single concept. Such an approach implies to turn back expertise against itself. Copyright © 2017 ASAC. Published by John Wiley & Sons, Ltd.     

Incorporating active adjustment into a financing based model of capital structure

The conventional partial adjustment model, which focuses on leverage evolution, has difficulty identifying deliberate capital structure adjustments as it confounds financing decisions with the mechanical autocorrelation of leverage. We propose and estimate a financing-based partial adjustment model that separates the effects of financing decisions on leverage evolution from mechanical evolution. The speed of adjustment (SOA) is firm-specific and stochastic, and active targeting of capital structure has a multiplier effect that depends on the size of financial deficit. Overall, we find expected SOA from active rebalancing (30%) more than doubles what is expected from mechanical mean reversion alone (13%).     

Entry in contests with incomplete information: Theory and experiments

This paper studies entry decisions in contests with private values. Potential contestants observe their value and the common opportunity cost of entry, and make entry decisions simultaneously. Theory predicts that whether or not contestants are informed of the number of entrants prior to choosing their expenditures has no effect on entry or aggregate expenditures. We test these assertions in our experiments. We find substantial over-entry in both information structures. However, entry is higher when contestants are informed. Since expenditures do not, on average, differ across information structures, aggregate expenditure is also higher when contestants are informed. Contestants earn on average less than the opportunity cost of entry.     

Models of static and dynamic technical efficiency of municipal libraries in the Czech Republic

This article deals with the modelling of the static and dynamic technical efficiency under conditions of municipal libraries of municipalities with 1000–5000 inhabitants. The aim of this article is to determine the level of the technical efficiency and the factors that influence the results of modelling of the static and dynamic technical efficiency of 34 selected municipal libraries for the years of 2011 and 2015. The first model tests the technical efficiency of conventional services of public libraries. The second model tests the technical efficiency of municipal libraries’ operation. The third model tests the technical efficiency of the key revenues and expenditures. The results in the static models estimate the average technical efficiency of municipal libraries in the interval (0.691–0.759) for the input-oriented models, and in the interval (1.413–2.005) for the output-oriented models. In the dynamic models, the majority of municipal libraries in 2015 showed lower technical efficiency and productivity in comparison with the year of 2011. The factors influencing the level of efficiency and its course include the inputs and the outputs, and their combinations within individual models.     

The effect of mandatory CSR disclosure on firm profitability and social externalities: Evidence from China

We examine how mandatory disclosure of corporate social responsibility (CSR) impacts firm performance and social externalities. Our analysis exploits China's 2008 mandate requiring firms to disclose CSR activities, using a difference-in-differences design. Although the mandate does not require firms to spend on CSR, we find that mandatory CSR reporting firms experience a decrease in profitability subsequent to the mandate. In addition, the cities most impacted by the disclosure mandate experience a decrease in their industrial wastewater and SO₂ emission levels. These findings suggest that mandatory CSR disclosure alters firm behavior and generates positive externalities at the expense of shareholders.