Article
Multidisciplinary Sciences
Joydeb Kumar Sana, Mohammad Zoynul Abedin, M. Sohel Rahman, M. Saifur Rahman
Summary: Customer churn is a critical issue in the telecommunications industry. We have investigated machine learning models and data transformation methods to improve the accuracy of customer churn prediction. Through feature selection and hyperparameter optimization, we have demonstrated the effectiveness of these techniques in improving prediction performance.
Article
Mathematics, Applied
Bin Li, Yuan Lei
Summary: Inspired by Dax's hybrid algorithm, this study establishes three strategies for solving linear inequalities and designs three different formulas based on the identification property. A distinctive feature of the switching strategies is the adaptive estimates of the optimal active set, which is shown to be efficient in extensive numerical experiments, especially for inconsistent systems.
NUMERICAL ALGORITHMS
(2022)
Article
Biochemistry & Molecular Biology
Claudio S. Quilodran, Mathias Currat, Juan I. Montoya-Burgos
Summary: Testing associations between objects is crucial in ecology, evolution, and quantitative sciences. This study evaluates the statistical power and error rate of different types of variables in the presence of spatial autocorrelation, and provides guidelines for choosing appropriate methods.
MOLECULAR ECOLOGY RESOURCES
(2023)
Review
Computer Science, Interdisciplinary Applications
Stephane C. K. Tekouabou, Stefan Cristian Gherghina, Eric Desire Kameni, Youssef Filali, Khalil Idrissi Gartoumi
Summary: The advancement of urban digitization has led to the emergence of AI-based tools in decision support systems. These tools use machine learning algorithms to extract valuable information, such as house price predictions. The urban real estate investment business model is undergoing a fundamental change with the support of AI data analysis. However, there is a lack of evaluation on the existing research in this field, and future studies should focus on big data and model explainability challenges.
ARCHIVES OF COMPUTATIONAL METHODS IN ENGINEERING
(2023)
Article
Agronomy
Yingjie Lu, Tao Li, Hui Hu, Xuemei Zeng
Summary: This study established hybrid prediction models based on VMD, EMD, and EEMD combined with BPNN to improve the prediction accuracy of ET0 time series. The results indicated the superiority of the VMD-BPNN model in terms of accuracy and stability. The effectiveness of the VMD method in reducing the non-stationarity of the original daily ET0 data was demonstrated.
AGRICULTURAL WATER MANAGEMENT
(2023)
Article
Energy & Fuels
Thomas Lickert, Stefanie Fischer, James L. Young, Selina Klose, Irene Franzetti, Daniel Hahn, Zhenye Kang, Meital Shviro, Fabian Scheepers, Marcelo Carmo, Tom Smolinka, Guido Bender, Sebastian Metz
Summary: While the number of publications in the PEM water electrolysis community increases, there is still no common ground in terms of reference hardware and testing protocols. However, the Task 30 Electrolysis within the AFC TCP of IEA has made initial attempts in this area and identified improvements in test hardware and measurement protocols. This paper presents a detailed testing protocol for commercial reference components using a laboratory test cell developed by Fraunhofer ISE, which was evaluated at three different institutions simultaneously. The importance of temperature control and conditioning phase is also demonstrated.
Article
Biochemical Research Methods
Bin Li, Wen Zhang, Chuang Guo, Hao Xu, Longfei Li, Minghao Fang, Yinlei Hu, Xinye Zhang, Xinfeng Yao, Meifang Tang, Ke Liu, Xuetong Zhao, Jun Lin, Linzhao Cheng, Falai Chen, Tian Xue, Kun Qu
Summary: This study compares 16 integration methods using 45 paired datasets and 32 simulated datasets, finding that Tangram, gimVI, and SpaGE perform best in predicting RNA transcript distribution, while Cell2location, SpatialDWLS, and RCTD are top-performing methods for cell type deconvolution. The benchmark pipeline provided in the study helps researchers in selecting optimal integration methods for their datasets. The comprehensive benchmarking analysis presented in this work evaluates computational methods integrating spatial and single-cell transcriptomics data for transcript distribution prediction and cell type deconvolution.
Article
Chemistry, Physical
Matus Dubecky, Stanislav Minarik, Frantisek Karlicky
Summary: In this study, we provide a benchmark for the fundamental gap of Sc2C(OH)(2), a non-magnetic member of MXenes, using many-body GW and fixed-node diffusion Monte Carlo methods. Both methods yield a similar value of Delta around 1.3 eV, validating their accuracy. The bandgap of Sc2C(OH)(2) makes it a suitable 2D semiconductor for optoelectronic applications.
JOURNAL OF CHEMICAL PHYSICS
(2023)
Article
Psychology
Roberta Rocca, Tal Yarkoni
Summary: Consensus on common evaluation benchmarks is lacking in the field of psychology, leading to idiosyncratic evaluation practices and difficulty in discerning between good and bad models. Introducing common benchmarks is advocated as a means to drive cumulative progress in psychology, emphasizing the practical utility of scientific models.
ADVANCES IN METHODS AND PRACTICES IN PSYCHOLOGICAL SCIENCE
(2021)
Article
Chemistry, Physical
Pierre-Paul De Breuck, Geoffroy Hautier, Gian-Marco Rignanese
Summary: The MODNet framework, utilizing a feedforward neural network and selecting physically meaningful features, outperforms current graph-network models on small datasets by accurately predicting material properties and enabling the prediction of multiple properties at once.
NPJ COMPUTATIONAL MATERIALS
(2021)
Article
Mathematics
Salam Salameh Shreem, Mohd Zakree Ahmad Nazri, Salwani Abdullah, Nor Samsiah Sani
Summary: Selecting the most minimal set of genes from microarray datasets for clinical diagnosis and prediction is a challenging task in machine learning. This study proposes a gene selection method called SU-RSHSA that combines the advantages of the Symmetrical Uncertainty (SU) filter and Reference Set Harmony Search Algorithm (RSHSA) wrapper to generate a small subset of genes with high classification accuracy.
Article
Multidisciplinary Sciences
Yasmine S. Al-Hamdani, Peter R. Nagy, Andrea Zen, Dennis Barton, Mihaly Kallay, Jan Gerit Brandenburg, Alexandre Tkatchenko
Summary: Quantum-mechanical methods are widely employed for describing molecular interactions, but discrepancies between CCSD(T) and DMC interaction energies for a set of polarizable supramolecules call for further collaborative efforts to resolve this issue.
NATURE COMMUNICATIONS
(2021)
Article
Chemistry, Physical
Qiaohao Liang, Aldair E. Gongora, Zekun Ren, Armi Tiihonen, Zhe Liu, Shijing Sun, James R. Deneault, Daniil Bash, Flore Mekki-Berrada, Saif A. Khan, Kedar Hippalgaonkar, Benji Maruyama, Keith A. Brown, John Fisher Iii, Tonio Buonassisi
Summary: Bayesian optimization (BO) has shown effectiveness in guiding autonomous and high-throughput experiments in materials science. This study evaluated the efficiency of BO across five different experimental materials systems, finding that Gaussian Process (GP) with anisotropic kernels and Random Forest (RF) performed well in BO, outperforming commonly used GP with isotropic kernels. GP with anisotropic kernels demonstrated robustness, while RF is a close alternative with advantages such as lack of distribution assumptions, lower time complexity, and easier initial hyperparameter selection. The benefits of using GP with anisotropic kernels in future materials optimization campaigns were also highlighted.
NPJ COMPUTATIONAL MATERIALS
(2021)
Article
Computer Science, Interdisciplinary Applications
Xifu Sun, Barry Croke, Anthony Jakeman, Stephen Roberts
Summary: Active Subspaces is a recently developed concept that identifies essential directions of the response surface, providing sensitivity metrics known as activity scores. Comparisons with other global methods show that activity scoring is accurate and computationally efficient in most cases.
ENVIRONMENTAL MODELLING & SOFTWARE
(2022)
Article
Chemistry, Multidisciplinary
Raphael Mazzine Barbosa de Oliveira, David Martens
Summary: The study conducted a large benchmarking study on the performance of ten counterfactual generation algorithms on 22 datasets, and introduced a novel framework for testing these algorithms. The results show that there is no single best algorithm for generating counterfactual explanations, as performance highly depends on specific properties of the dataset and model.
APPLIED SCIENCES-BASEL
(2021)
Article
Chemistry, Inorganic & Nuclear
Hillary Pan, Alex M. Ganose, Matthew Horton, Muratahan Aykol, Kristin A. Persson, Nils E. R. Zimmermann, Anubhav Jain
Summary: This study introduces the MaterialsCoord benchmark suite and CrystalNN near-neighbor algorithm, experimentally comparing CrystalNN with other algorithms and finding its performance to be similar to well-established ones. Additionally, computational demand and sensitivity to thermal motion simulations were evaluated for each algorithm.
INORGANIC CHEMISTRY
(2021)
Correction
Chemistry, Inorganic & Nuclear
Hillary Pan, Alex M. Ganose, Matthew Horton, Muratahan Aykol, Kristin Persson, Nils E. R. Zimmermann, Anubhav Jain
INORGANIC CHEMISTRY
(2021)
Article
Multidisciplinary Sciences
Alex M. Ganose, Junsoo Park, Alireza Faghaninia, Rachel Woods-Robinson, Kristin A. Persson, Anubhav Jain
Summary: The authors developed a computationally efficient method for calculating carrier scattering rates of semiconductors, which shows similar accuracy to state-of-the-art methods but at a much lower computational cost. This approach enables high-throughput computational workflows for accurate screening of carrier mobilities, lifetimes, and thermoelectric power.
NATURE COMMUNICATIONS
(2021)
Article
Materials Science, Multidisciplinary
Xuemin Shi, Xinyue Zhang, Alexander Ganose, Junsoo Park, Cheng Sun, Zhiwei Chen, Siqi Lin, Wen Li, Anubhav Jain, Yanzhong Pei
Summary: The n-type Mg3Sb2-based materials have shown promising potential for thermoelectric applications due to their high band degeneracy, high carrier mobility, low lattice thermal conductivity, and environmental advantages. The addition of Mg3Bi2 alloy significantly improves their performance, but the optimal alloying concentrations vary with temperature due to a compromise between band structure and phonon scattering effects.
MATERIALS TODAY PHYSICS
(2021)
Article
Multidisciplinary Sciences
Drew Lilley, Peiyuan Yu, Jason Ma, Anubhav Jain, Ravi Prasher
Summary: Thermal fluids play a crucial role in energy technologies, but current thermal fluids still have lower heat capacity than water and higher viscosity. The proposed thermochemical fluid has significantly higher heat capacity over a wide temperature range, lower viscosity than water, and a model is developed for designing future thermal fluids with even higher heat capacity.
Article
Multidisciplinary Sciences
Alex M. Ganose, David O. Scanlon, Aron Walsh, Robert L. Z. Hoye
Summary: The optoelectronic performance of wide-bandgap semiconductors is often inferior to that of their defect-tolerant small-bandgap counterparts. The authors highlight three main challenges that need to be addressed in order to mitigate the impact of defects in wide-bandgap semiconductors.
NATURE COMMUNICATIONS
(2022)
Article
Chemistry, Physical
Xinwei Wang, Alex M. Ganose, Sean R. Kavanagh, Aron Walsh
Summary: Antimony sulfide and selenide have emerged as promising absorbers for photovoltaic applications, with band-like charge-carrier transport governed by large polarons. The room-temperature carrier mobility is limited by scattering from polar phonon modes and sample defects.
ACS ENERGY LETTERS
(2022)
Article
Chemistry, Physical
Adam J. Jackson, Benjamin J. Parrett, Joe Willis, Alex M. Ganose, W. W. Winnie Leung, Yuhan Liu, Benjamin A. D. Williamson, Timur K. Kim, Moritz Hoesch, Larissa S. I. Veiga, Raman Kalra, Jens Neu, Charles A. Schmuttenmaer, Tien-Lin Lee, Anna Regoutz, Tung-Chun Lee, Tim D. Veal, Robert G. Palgrave, Robin Perry, David O. Scanlon
Summary: This study demonstrates the ideal transparent conducting oxide ZnSb2O6 and identifies gallium as the optimal dopant for achieving high conductivity and transparency. Experimental validation of computational predictions shows that Ga-doped ZnSb2O6 exhibits behavior consistent with a degenerate transparent conducting oxide.
ACS ENERGY LETTERS
(2022)
Article
Chemistry, Inorganic & Nuclear
Guikai Zheng, Qi Wang, Jinfan Chen, Ruizhi Qiu, Min Zhu
Summary: This study uses density functional theory to investigate the local cluster structures, energetic stabilities, and electronic properties of UO2 grain boundaries (GBs). The presence of new medium-range U-O bonds in the GBs is highlighted, which is often overlooked in conventional coordination analysis. The Smooth Overlap of Atomic Positions (SOAP) method is used to quantitatively describe the distorted clusters, and a dissimilarity index is defined to measure the differences between the distorted and perfect structures. The results show that the medium-range SOAP dissimilarity is well correlated with the GB excess energy, and high-energy GBs have shortened band gaps primarily contributed by distorted U clusters.
INORGANIC CHEMISTRY
(2023)
Article
Chemistry, Physical
Ruiqi Wu, Alex M. Ganose
Summary: The study discovers that K2CsSb exhibits ideal properties for use in photovoltaic applications, and its maximum theoretical efficiency can compete with other state-of-the-art photovoltaics.
JOURNAL OF MATERIALS CHEMISTRY A
(2023)
Article
Materials Science, Multidisciplinary
Rachel Woods-Robinson, Yihuang Xiong, Jimmy-Xuan Shen, Nicholas Winner, Matthew K. Horton, Mark Asta, Alex M. Ganose, Geoffroy Hautier, Kristin A. Persson
Summary: Many semiconductors have weak or forbidden transitions at their band edges, creating a widened transparency region. The presence of forbidden transitions has been neglected in the study of new p-type transparent conductors, unlike in n-type transparent conductors. Through computation and analysis of absorption spectra, it is found that compounds with highly localized band-edge states are more likely to exhibit forbidden transitions. This study suggests potential p-type and n-type transparent conductor candidates with forbidden transitions.
Article
Chemistry, Physical
Ju Huang, Matthias J. Golomb, Sean R. Kavanagh, Kasper Tolborg, Alex M. Ganose, Aron Walsh
Summary: Covalent organic frameworks (COFs) exhibit a high degree of flexibility in both chemical and structural aspects, with a diverse range of stacking sequences accessible. Computational study reveals a preference for slipped structures with horizontal offsets in two prototypical COFs, leading to significant changes in the underlying electronic structure and a band gap opening of 0.8-1.4 eV. Implications for screening and selecting COFs for energy applications are discussed.
JOURNAL OF MATERIALS CHEMISTRY A
(2022)
Article
Computer Science, Artificial Intelligence
Amalie Trewartha, Nicholas Walker, Haoyan Huo, Sanghoon Lee, Kevin Cruse, John Dagdelen, Alexander Dunn, Kristin A. Persson, Gerbrand Ceder, Anubhav Jain
Summary: Efficiently connecting new materials discoveries to established literature can be achieved by using named entity recognition (NER) to extract structured summary-level data from unstructured materials science text. In this study, the performance of four NER models on materials science datasets is compared, and it is found that domain-specific pre-training provides measurable advantages, with the bidirectional long short-term memory (BiLSTM) model consistently outperforming BERT.
Article
Chemistry, Physical
Xinwei Wang, Zhenzhu Li, Sean R. Kavanagh, Alex M. Ganose, Aron Walsh
Summary: Antimony sulfide and selenide have emerged as promising alternatives for thin-film photovoltaic compounds. The anisotropic crystal structures of these materials, composed of quasi-one-dimensional ribbons, have been found to exhibit inter-ribbon interactions beyond the van der Waals regime. The stereochemical activity of the lone pair of Sb 5s is shown to be the origin of the anisotropic structures. The impacts of structural anisotropy on the electronic, dielectric, and optical properties, as well as charge carrier transport, are examined, providing guidelines for optimizing the performance of Sb2X3-based photovoltaics.
PHYSICAL CHEMISTRY CHEMICAL PHYSICS
(2022)