Figure - available from: Annals of Operations Research
This content is subject to copyright. Terms and conditions apply.
CART grown on polar coordinates (top) and corresponding court partition induced by CART (bottom)—Stephen Curry, NBA regular season 2020/2021
Source publication
This paper proposes tools for spatial performance analysis in basketball. In detail, we aim at representing maps of the court visualizing areas with different levels of scoring probability of the analysed player or team. To do that, we propose the adoption of algorithmic modeling techniques. Firstly, following previous studies, we examine CART, hig...
Similar publications
Detection of nodes disseminating false data is a prerequisite for effective deployment of Internet of Vehicles (IoV)
services. This work proposed a novel hyper-tuned ensemble Random Forest (Ens. RF) algorithm to detect false basic
safety messages in IoV. Performance evaluation done using the Vehicular Reference Misbehavior (VeReMi) dataset
comprisi...
Citations
... Data analysis using the Classification and Regression Tree (CART) instrument was a follow-up analysis of Chi-square analysis to find the explanatory variables in stages through the pruning stages (Shabani 2017;Zuccolotto et al. 2023). The response variable in the CART analysis was the rate of forest encroachment and wood theft (Y) by people living around the forest areas, where the explanatory variable was a natural disaster (Xn), as presented in Table 1. ...
Lore Lindu National Park (LLNP) is a conservation area that contains a lot of wood resources. Various illegal community activities have become widespread, such as illegal mining and illegal logging. So, this research aims to determine the involvement of communities around forest areas in material and wood theft from June to October 2021. To determine forest encroachment, we find explanatory variables, using qualitative description integrated with perceptual tests and Classification and Regression Tree (CART) analysis. Based on the results of the 10-fold cross-validation analysis with the smallest Rcv (x-Val relative error) value of 0.428, with a classification accuracy of 68.6%, a four-node optimum tree was obtained, which explained that as many as 86 forest encroachers were victims of a vast landslide disaster along with flood and whirlwind, due to which there was no longer any property left for them. Their encroachment affected the condition of land cover. The data on the land cover change, from 2010 to 2020, showed a reduction of 15,369.20 ha or 6.90%, which indicated a severe threat to the sustainability of LLNP as a biodiversity conservation area that should be protected. The involvement in illegal logging by communities living around the forest areas resulted from the loss of their agricultural land for their livelihoods due to natural disasters such as flood, landslide and whirlwind that destroyed infrastructure and community settlement facilities. As a result, these losses and destruction were a catalyst for forest destruction. Initially being in the frontline for preserving the forest, however, the community has now turned into silent partners with licensed wood businesspeople. The community eventually becomes a subsystem in the social ecology system (SES), which negatively affects the destruction of forest resources, production and conservation forests.
With the advancements in artificial intelligence technology, research on predicting match results using machine learning is actively conducted in various sports. However, there is lack of research on match prediction in professional football, a sport with the largest industrial scale. To overcome this limitation, this study collected data from 762 matches held in the Korean Professional Football League from February 25, 2020, to July 22, 2023, through the K-League Data Portal. Using Python 3.10.9, six machine learning algorithms were utilized to predict match results. The analysis revealed that, based on linear regression, the results of 119 matches(82.6%) of a total of 144 matches were predicted accurately. This study holds theoretical and practical significance as a pioneering empirical study in South Korea that applies machine learning to predict match results in the Korean Professional Football League.
Transparent and explainable machine learning (ML) models are essential in various domains, e.g., energy consumption, where decarbonization is the main challenge. The European Union is focusing on energy efficiency retrofits in residential buildings to help reach its 2050 carbon emissions target. The cost of these investments is often a strong factor, requiring decision-makers to understand the motivations driving homeowners’ decisions to undertake energy retrofits. Instead of hedonic models commonly used in operational management research studies, we rely on ML methods to predict homeowners’ decisions to undertake energy retrofits, using data from 51,000 households in France. We describe the data preparation, model training, and evaluation; results show that artificial neural networks outperform other popular ML techniques (91.4%). Our post hoc method based on sensitivity analysis and feature importance contributes to the transparency and interpretability of the results. We show that the type of public aid used, head of household gender, family size, prior knowledge of aid, urban vs. rural area, geographical location, occupancy status, and working status are the most important factors in the decision to undertake energy efficiency retrofits. Our predictive methods help decision-makers to make optimal decisions about the level, type, beneficiaries of public incentives for energy retrofits, and expected outcomes; companies in the construction sector can understand homeowners’ key motivations and optimally calibrate their strategic investments and operations.
