Model Validation Techniques Research Articles

Using artificial intelligence based imaging to predict lymph node metastasis in non-small cell lung cancer: a systematic review and meta-analysis.

Lung cancer, especially non-small cell lung cancer (NSCLC), is one of the most-deadly malignancies worldwide. Lung cancer has a worse 5-year survival rate than many primary malignancies. Thus, the early detection and prognosis prediction of lung cancer are crucial. The early detection and prognosis prediction of lung cancer have improved with the widespread use of artificial intelligence (AI) technologies. This meta-analysis examined the accuracy and efficacy of AI-based models in predicting lymph node metastasis (LNM) in NSCLC patients using imaging data. Our findings could help clinicians predict patient prognosis and select alternative therapies. We searched the PubMed, Web of Science, Cochrane Library, and Embase databases for relevant articles published up to January 31, 2024. Two reviewers individually evaluated all the retrieved articles to assess their eligibility for inclusion in the meta-analysis. The systematic assessment and meta-analysis comprised articles that satisfied the inclusion criteria (e.g., randomized or non-randomized trials, and observational studies) and exclusion criteria (e.g., articles not published in English), and provided data for the quantitative synthesis. The quality of the included articles was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2). The pooled sensitivity, specificity, and area under the curve (AUC) were used to evaluate the ability of AI-based imaging models to predict LNM in NSCLC patients. Sources of heterogeneity were investigated using meta-regression. Covariates, including country, sample size, imaging modality, model validation technique, and model algorithm, were examined in the subgroup analysis. The final meta-analysis comprised 11 retrospective studies of 6,088 NSCLC patients, of whom 1,483 had LNM. The pooled sensitivity, specificity, and AUC of the AI-based imaging model for predicting LNM in NSCLC patients were 0.87 [95% confidence interval (CI): 0.80-0.91], 0.85 (95% CI: 0.78-0.89), and 0.92 (95% CI: 0.90-0.94). Based on the QUADAS-2 results, a risk of bias was detected in the patient selection and diagnostic tests of the included articles. However, the quality of the included articles was generally acceptable. The pooled sensitivity and specificity were heterogeneous (I2>75%). The meta-regression and subgroup analyses showed that imaging modality [computed tomography (CT) or positron emission tomography (PET)/CT], and the neural network method model design significantly affected heterogeneity of this study. Models employing sample size data from a single center and the least absolute shrinkage and selection operator (LASSO) method had greater sensitivity than other techniques. Using the Deek' s funnel plot, no publishing bias was found. The results of the sensitivity analysis showed that deleting each article one by one did not change the findings. Imaging data models based on AI algorithms have good diagnostic accuracy in predicting LNM in patients with NSCLC and could be applied in clinical settings.

On the Use of SuperDARN Ground Backscatter Measurements for Ionospheric Propagation Model Validation

AbstractPrior to use in operational systems, it is essential to validate ionospheric models in a manner relevant to their intended application to ensure satisfactory performance. For Over‐the‐Horizon radars (OTHR) operating in the high‐frequency (HF) band (3–30 MHz), the problem of model validation is severe when used in Coordinate Registration (CR) and Frequency Management Systems (FMS). It is imperative that the full error characteristics of models is well understood in these applications due to the critical relationship they impose on system performance. To better understand model performance in the context of OTHR, we introduce an ionospheric model validation technique using the oblique ground backscatter measurements in soundings from the Super Dual Auroral Radar Network (SuperDARN). Analysis is performed in terms of the F‐region leading edge (LE) errors and assessment of range‐elevation distributions using calibrated interferometer data. This technique is demonstrated by validating the International Reference Ionosphere (IRI) 2016 for January and June in both 2014 and 2018. LE RMS errors of 100–400 km and 400–800 km are observed for winter and summer months, respectively. Evening errors regularly exceeding 1,000 km across all months are identified. Ionosonde driven corrections to the IRI‐2016 peak parameters provide improvements of 200–800 km to the LE, with the greatest improvements observed during the nighttime. Diagnostics of echo distributions indicate consistent underestimates in model NmF2 during the daytime hours of June 2014 due to offsets of −8° being observed in modeled elevation angles at 18:00 and 21:00 UT.

Machine Learning in Coastal Engineering: Applications, Challenges, and Perspectives

The integration of machine learning (ML) techniques in coastal engineering marks a paradigm shift in how coastal processes are modeled and understood. While traditional empirical and numerical models have been stalwarts in simulating coastal phenomena, the burgeoning complexity and computational demands have paved the way for data-driven approaches to take center stage. This review underscores the increasing preference for ML methods in coastal engineering, particularly in predictive tasks like wave pattern prediction, water level fluctuation, and morphology change. Although the scope of this review is not exhaustive, it aims to spotlight recent advancements and the capacity of ML techniques to harness vast datasets for more efficient and cost-effective simulations of coastal dynamics. However, challenges persist, including issues related to data availability and quality, algorithm selection, and model generalization. This entails addressing fundamental questions about data quantity and quality, determining optimal methodologies for specific problems, and refining techniques for model training and validation. The reviewed literature paints a promising picture of a future where ML not only complements but significantly enhances our ability to predict and manage the intricate dynamics of coastal environments.

Analysis of age-specific fertility in India: Deterministic and non-deterministic modeling approaches

The main objective of this study is to investigate the pattern of age-specific fertility rates (ASFRs) in India using deterministic and non-deterministic approaches. Toward this end, we proposed statistical polynomial regression models to study the distributional pattern of ASFRs for total, rural, and urban women in India. Further, a comparative study considering selected skewed regression models was undertaken. For this study, secondary data on ASFR were collected from Sample Registration System, Statistical Report-2020, and from National Family Health Survey 5 (NFHS-5; 2019 – 2021). It was found that all three subcategories of ASFRs, namely, the total, rural, and urban ASFRs, followed the reciprocal biquadratic polynomial model. On the other hand, all three subcategories of ASFR follow the skew-normal type 2 distribution. Similar findings were also obtained and validated based on NFHS-5 data. Further, the chosen statistical models’ validity and stability were tested using various model validation techniques and model selection criteria.

Quality indices for topic model selection and evaluation: a literature review and case study

BackgroundTopic models are a class of unsupervised machine learning models, which facilitate summarization, browsing and retrieval from large unstructured document collections. This study reviews several methods for assessing the quality of unsupervised topic models estimated using non-negative matrix factorization. Techniques for topic model validation have been developed across disparate fields. We synthesize this literature, discuss the advantages and disadvantages of different techniques for topic model validation, and illustrate their usefulness for guiding model selection on a large clinical text corpus.Design, setting and dataUsing a retrospective cohort design, we curated a text corpus containing 382,666 clinical notes collected between 01/01/2017 through 12/31/2020 from primary care electronic medical records in Toronto Canada.MethodsSeveral topic model quality metrics have been proposed to assess different aspects of model fit. We explored the following metrics: reconstruction error, topic coherence, rank biased overlap, Kendall’s weighted tau, partition coefficient, partition entropy and the Xie-Beni statistic. Depending on context, cross-validation and/or bootstrap stability analysis were used to estimate these metrics on our corpus.ResultsCross-validated reconstruction error favored large topic models (K ≥ 100 topics) on our corpus. Stability analysis using topic coherence and the Xie-Beni statistic also favored large models (K = 100 topics). Rank biased overlap and Kendall’s weighted tau favored small models (K = 5 topics). Few model evaluation metrics suggested mid-sized topic models (25 ≤ K ≤ 75) as being optimal. However, human judgement suggested that mid-sized topic models produced expressive low-dimensional summarizations of the corpus.ConclusionsTopic model quality indices are transparent quantitative tools for guiding model selection and evaluation. Our empirical illustration demonstrated that different topic model quality indices favor models of different complexity; and may not select models aligning with human judgment. This suggests that different metrics capture different aspects of model goodness of fit. A combination of topic model quality indices, coupled with human validation, may be useful in appraising unsupervised topic models.

Misuse of multinomial logistic regression in stroke related health research: A systematic review of methodology.

Multinomial logistic regression (MLR) is often used to model the association between a nominal outcome variable and one or more covariates. The results of MLR are interpreted as relative risk ratios (RRR) and warrant a more coherent interpretation than ordinary logistic regression. Some authors compare the results of MLR to ordinal logistic regression (OLR), irrespective of the fact that these estimate different quantities. We aim to investigate the time trends in the use and misuse of MLR in studies including stroke patients, specifically the extent to which (1) the results are denoted as anything other than RRR, (2) comparisons are made of results with results of OLR and (3) results have been interpreted coherently. Secondarily, we examine the use of model validation techniques in studies with predictive aims. We searched EMBASE and PubMed for articles using MLR on populations of stroke patients. Identified studies were screened, and information pertaining to our aims was extracted. A total of 285 articles were identified through a systematic literature search, and 68 of these were included in the review. Of these, 60 articles (88%) did not denote exponentiated coefficients of MLR as relative risk ratios but rather some other measure. Additionally, 63 articles (93%) interpreted the results of MLR in a non-coherent manner. Two articles attempted to compare MLR results with those of OLR. Nine studies attempted to use MLR for predictive means, and three used relevant validation techniques. From these findings, it is clear that the interpretation of MLR is often suboptimal.

Application of machine learning algorithms to predict osteoporosis in postmenopausal women with type 2 diabetes mellitus.

The screening and diagnosis of osteoporosis in patients with type 2 diabetes mellitus (T2DM) based on bone mineral density remains challenging because of the limited availability and accessibility of dual-energy X-ray absorptiometry. We aimed to develop and validate models to predict the risk of osteoporosis in postmenopausal women with T2DM based on machine learning (ML) algorithms. This retrospective study included 303 postmenopausal women with T2DM. To develop prediction models for osteoporosis, we applied nine ML algorithms combined with demographic, clinical, and laboratory data. The least absolute shrinkage and selection operator were used to perform feature selection. We used the bootstrap resampling technique for model training and validation. To test the performance of the models, we calculated indices including the area under the receiver operating characteristic curve (AUROC), accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, calibration curve, and decision curve analysis. Furthermore, we conducted fivefold cross-validation for parameter optimization and model validation. Feature importance was assessed using the SHapley additive explanation (SHAP). We identified 10 independent predictors as the most valuable features. An AUROC of 0.616-1.000 was observed for nine ML algorithms. The extreme gradient boosting (XGBoost) model exhibited the best performance, outperforming conventional risk assessment tools and registering 0.993 in the training set, 0.798 in the validation set, and 0.786 in the test set for fivefold cross-validation. Using SHAP, we found that the explanatory variables contributed to the model and their relationship with osteoporosis occurrence. Furthermore, we developed a user-friendly tool for calculating the risk of osteoporosis. With the integration of demographic and clinical risk factors, ML algorithms can accurately predict osteoporosis. The XGBoost model showed ideal performance. With the incorporation of these models in the clinic, patients may benefit from early osteoporosis diagnosis and treatment.

ASSESSING LONG-TERM IMPACTS OF DISASTER USING PREDICTIVE DATA ANALYTICSFOR EFFECTIVE DECISION SUPPORT

Disaster is a big issue that seriously disrupts and affects the community or society. The impact of a disaster causes a short and long period of time. To analyze the impacts of disasters there are lots of available related datasets. Data analytics methods have the potential to assess the impacts of different types of disasters. Collectively data analytics and machine learning techniques play an important role in transforming and being able to make decisions about our social, economic, mental, and psychological things. The objective of this paper is to assess the impacts of disasters from immediate term to long-term, provide crucial help to the emergency management workforce, and policy decisions making based on the latest available datasets. With the help of the various data agencies, extraction of information and activities carried out, we can determine the effects on disaster victims, their community and impacts on society in general. The analysis provides the statistics that can guide our emergency service about the status of facilities that can further support the survivors, and other related information. Detailed assessment i.e., structural survey and hazard mapping provide specific information about reconstruction and mitigation to monitor the situation, needs of the victims, and supporting entities. The assessment is based on the type of disaster that happened and its impact after a few years. In the current technological advancement of data analytics and machine learning algorithms, the prediction of the long-term effects of a disaster can be performed. Analyzing the impacts over a long period of time is also dependent on the growth of actively cared datasets gathering bodies like agencies, government, NGOs, media, etc., where prediction of short-term and long-term impacts is dependent on the available datasets. Available datasets are preprocessed using data analytics tools and implementation of training and testing for the purpose of predictions and recommendations. As a huge amount of data sets are available through different sources the classification of the datasets can also be performed resulting fast and accurate processing. Model validation techniques play an important role to check the validation, test result, and related outcomes. In this paper advanced machine learning and data analytics tools i.e., XG boost, modified SVM, and modified RF are used for better prediction. The analysis of the short-term effects of disasters has already been suggested and recommended by the various conventional approaches. Here the focus is to analyze and detect the long-term effects of a disaster along with recommendations and models preparing for good decision-making. Therefore, planning should be focused on assessing the impacts from short-term to long-term. The findings of the paper would be helpful to the agencies, local & national authorities, and the government by recommending action plans and their future effects for a longer period in case of disaster.

Sonic Well-Log Imputation Through Machine-Learning-Based Uncertainty Models

Sonic well logs provide critical information to calibrate seismic data and support geomechanical characterization. Advanced subsurface data analytics and machine learning enable new methods and workflows for property estimation, regression, and classification for geoscience and subsurface engineering applications. However, current applications for imputation of well-logging values rely only on model accuracy and low error predictions. T raditional model validation techniques are not enough to validate models and account for the substantial uncertainty in the subsurface. Well-logging imputation estimates and their associated uncertainty models are essential to the field development planning and decision-making workflows, such as reservoir modeling, volumetric resource assessment, predrill prediction with uncertainty, remaining resource mapping, and production allocation. When performing subsurface feature imputation with machine learning, we must expand our machine-learning model training and complexity tuning workflows to check the entire uncertainty model to ensure uncertainty distributions are precise and accurate. We propose a workflow that integrates the goodness metric to calculate accurate and precise uncertainty models of sonic well-log predictions based on ensembles of the machine-learning estimates. Our workflow combines model evaluation and visualization of the estimates and the uncertainty model with respect to measured depth. Our proposed method provides intuitive diagnostics and metrics to evaluate estimation accuracy and uncertainty model goodness.

Machine learning analysis of confounding variables of a convolutional neural network specific for abdominal aortic aneurysms

ObjectiveTo identify confounding variables influencing the accuracy of a convolutional neural network (CNN) specific for infrarenal abdominal aortic aneurysms (AAAs) on computed tomography angiograms (CTAs). MethodsA Health Insurance Portability and Accountability Act-compliant, institutional review board-approved, retrospective study analyzed abdominopelvic CTA scans from 200 patients with infrarenal AAAs and 200 propensity-matched control patients. An AAA-specific trained CNN was developed by the application of transfer learning to the VGG-16 base model using model training, validation, and testing techniques. Model accuracy and area under the curve were analyzed based on data sets (selected, balanced, or unbalanced), aneurysm size, extra-abdominal extension, dissections, and mural thrombus. Misjudgments were analyzed by review of heatmaps, via gradient weighted class activation, overlaid on CTA images. ResultsThe trained custom CNN model reported high test group accuracies of 94.1%, 99.1%, and 99.6% and area under the curve of 0.9900, 0.9998, and 0.9993 in selected (n = 120), balanced (n = 3704), and unbalanced image sets (n = 31,899), respectively. Despite an eightfold difference between balanced and unbalanced image sets, the CNN model demonstrated high test group sensitivities (98.7% vs 98.9%) and specificities (99.7% vs 99.3%) in unbalanced and balanced image sets, respectively. For aneurysm size, the CNN model demonstrates decreasing misjudgments as aneurysm size increases: 47% (16/34) for aneurysms <3.3 cm, 32% (11/34) for aneurysms 3.3 to 5 cm, and 20% (7/34) for aneurysms >5 cm. Aneurysms containing measurable mural thrombus were over-represented within type II (false-negative) misjudgments compared with type I (false-positive) misjudgments (71% vs 15%, P < .05). Inclusion of extra-abdominal aneurysm extension (thoracic or iliac artery) or dissection flaps in these imaging sets did not decrease the model's overall accuracy, indicating that the model performance was excellent without the need to clean the data set of confounding or comorbid diagnoses. ConclusionsAnalysis of an AAA-specific CNN model can accurately screen and identify infrarenal AAAs on CTA despite varying pathology and quantitative data sets. The highest anatomic misjudgments were with small aneurysms (<3.3 cm) or the presence of mural thrombus. Accuracy of the CNN model is maintained despite the inclusion of extra-abdominal pathology and imbalanced data sets.

Spatial Validation of Agent-Based Models

This paper adapts an existing techno–social agent-based model (ABM) in order to develop a new framework for spatially validating ABMs. The ABM simulates citizen opposition to locally unwanted land uses, using historical data from an energy infrastructure siting process in Southern California. Spatial theory, as well as the model’s design, suggest that adequate validation requires multiple tests rather than relying solely on a single test-statistic. A pattern-oriented modeling approach was employed that first mapped real and simulated citizen comments across the US Census tract. The suite of spatial tests included Global Moran’s I, complemented with bivariate correlations, as well as the local indicators of spatial association (LISA) test. The global tests showed the model explained up to 65% of the variation in the historical data for US Census tract-level citizen comments on a locally unwanted land use. These global tests were also found helpful to inform the model’s calibration for the current application. The LISA results were even stronger, showing that the model predicted citizen comment clustering correctly in five of six Census tracts. It slightly over predicted comments further away from the land use. The LISA results and pattern-oriented modeling validation techniques identified theoretical factors to improve the modeling specification in future applications. The combined suite of validation techniques helped improve confidence in the model’s predictions.

Digitization of Prior Authorization in Healthcare Management Using Machine Learning

Background: Prior Authorization is a widely used process by health insurance companies in the United States before they agree to cover prescribed medication under medical insurance. How-ever, the traditional approach includes long-length papers, leading to patients' delayed processing of their claims. This delay may deteriorate the patient’s medical condition. Also, due to man-made er-rors, there is a chance of incorrect decision-making on the claims. On the other hand, physicians are losing their time getting their prescribed medication approved. It is essential to reduce the wait time of patients and the tedious work of physicians for healthcare to be effective. This demands advanced technology that can boost the decision-making process of prior authorization methodology. Objective: This work aims to digitize the prior authorization process by implementing classification algorithms to classify the initial authorization applications into Accepted/Rejected/Partially Accepted classes. A web application that inputs prior authorization claim details and outputs the predicted class of the claim was proposed. Methods: Analyzed and collected significant features by implementing feature selection. Developed classification models using Artificial Neural Networks and Random Forest. Implemented model val-idation techniques to evaluate classifier performance. Results: From the research findings, generic medication cost, type of health insurance plan, addictive nature and side effects of the prescribed drug, patient physical qualities like Age/Gender/Current Medical condition are the significant attributes that impact the decision-making process in the prior authorization process. Then, implemented classifiers exhibited accurate performance on the Train and Test data. Amongst Artificial Neural Networks classification model portrayed higher accuracy. Further a confusion matrix was further analyzed for developed models. In addition, k-fold cross-validation and availed performance evaluation metrics were conducted to validate the model perfor-mance. Conclusion: Ameliorated Healthcare by removing time and location barriers in the Prior Authoriza-tion process while ensuring patients get quality and economical medication. The proposed web ap-plication with a machine learning predictive model as a backend automates the prior authorization process by classifying the applications in a few seconds.

Study on the Natural Ventilation Characteristics of a Solar Greenhouse in a High-Altitude Area

The ventilation rate of a greenhouse is one of the major factors to consider when assessing its ventilation performance. Compared with plain areas, high-altitude areas have lower air pressure, thinner air, and stronger solar radiation, which in turn affect the magnitude of the local greenhouse ventilation rate. This paper is based on the use of online monitoring and computational fluid dynamics (CFD) techniques for modeling and model validation. The average relative error (ARE), mean absolute error (MAE), root-mean-square error (RMSE), and determination coefficient (R2) of the temperature were 4.88%, 1.396 °C, 1.428 °C, and 0.9982, respectively. The ARE, MAE, RMSE, and R2 of the velocity were 9.525%, 0.035 m/s, 0.049 m/s, and 0.9869, respectively. Then, the distributions of the wind pressure, Reynolds number (Re), thermal pressure, air density, air speed, and temperature in greenhouses in high-altitude and plain areas were researched to obtain the relevant factors affecting the ventilation rates of greenhouses in high-altitude areas. In addition, correlation analyses were conducted for five variables affecting the ventilation rate: the inlet velocity, the temperature difference between the inside and outside of the greenhouse, the air density difference between the inside and outside of the greenhouse, total indoor radiation, and the internal heat source of the crop, and the coefficients of their correlations with the greenhouse ventilation rate were 1.0, −0.83, −0.72, −0.72, and 0.68, respectively. A natural ventilation rate model for plateau areas was developed, with the ARE, RMSE, and R2 between the sample values and fitted values determined to be 4.55%, 0.543 m3/s, and 0.9997, respectively. The model was validated by predicting the greenhouse ventilation rate in winter (3 January 2022), and the ARE, RMSE, and R2 of the sample values and predicted values were 9.726%, 8.435 m3/s, and 0.9901, respectively. This study provides a theoretical basis for further research on greenhouse ventilation characteristics in high-altitude areas.

A Review on Rain Signal Attenuation Modeling, Analysis and Validation Techniques: Advances, Challenges and Future Direction

Radio waves are attenuated by atmospheric phenomena such as snow, rain, dust, clouds, and ice, which absorb radio signals. Signal attenuation becomes more severe at extremely high frequencies, usually above 10 GHz. In typical equatorial and tropical locations, rain attenuation is more prevalent. Some established research works have attempted to provide state-of-the-art reviews on modeling and analysis of rain attenuation in the context of extremely high frequencies. However, the existing review works conducted over three decades (1990 to 2022), have not adequately provided comprehensive taxonomies for each method of rain attenuation modeling to expose the trends and possible future research directions. Also, taxonomies of the methods of model validation and regional developmental efforts on rain attenuation modeling have not been explicitly highlighted in the literature. To address these gaps, this paper conducted an extensive literature survey on rain attenuation modeling, methods of analyses, and model validation techniques, leveraging the ITU-R regional categorizations. Specifically, taxonomies in different rain attenuation modeling and analysis areas are extensively discussed. Key findings from the detailed survey have shown that many open research questions, challenges, and applications could open up new research frontiers, leading to novel findings in rain attenuation. Finally, this study is expected to be reference material for the design and analysis of rain attenuation.

Forecasting the solar cycle 25 using a multistep Bayesian neural network

ABSTRACTPredicting the solar activity of upcoming cycles is crucial nowadays to anticipate potentially adverse space weather effects on the Earth’s environment produced by coronal transients and traveling interplanetary disturbances. The latest advances in deep learning techniques provide new paradigms to obtain effective prediction models that allow to forecast in detail the evolution of cosmogeophysical time series. Because of the underlying complexity of the dynamo mechanism in the solar interior that is at the origin of the solar cycle phenomenon, the predictions offered by state-of-the-art machine learning algorithms represent valuable tools for our understanding of the cycle progression. As a plus, Bayesian deep learning is particularly compelling thanks to recent advances in the field that provide improvements in both accuracy and uncertainty quantification compared to classical techniques. In this work, a deep learning long short-term memory model is employed to predict the complete profile of Solar Cycle 25, thus forecasting also the advent of the next solar minimum. A rigorous uncertainty estimation of the predicted sunspot number is obtained by applying a Bayesian approach. Two different model validation techniques, namely the Train-Test split and the time series k-fold cross-validation, have been implemented and compared, giving compatible results. The forecasted peak amplitude is lower than that of the preceding cycle. Solar Cycle 25 will last 10.6 ± 0.7 yr, reaching its maximum in the middle of the year 2024. The next solar minimum is predicted in 2030 and will be as deep as the previous one.

아파트 대량 평가를 위한 기계 학습 알고리즘과 스태킹 앙상블 기법의 응용

Purpose The goal is to find an optimal data-driven model for apartment price appraisal system. We compares the predictive power of several machine learning algorithms and the ensemble model by using 17,779 apartment transaction data in Gangnam from 2006 to 2017. Methods We use three tree-based machine learning techniques, random forest, Adaboost, and Catboost. Also, an ensemble model is developed through the stacking technique using the above algorithm. 5-fold cross validation is used as model validation techniques. Results By applying the algorithm, it was found that the ensemble model constructed from stacking method performs better than individual machine learning models. Conclusion The results illustrate that the quality of mass appraisals or apartment price indices can be improved by using the ensemble model.

Clustering and meta-envelopment in data envelopment analysis

We propose techniques of classification of a potentially heterogeneous data set into groups in a way that is consistent with the intended purpose of the clustering, which is Data Envelopment Analysis (DEA). Using standard clustering techniques and then applying DEA is shown to be sub-optimal in many instances of empirical relevance. Our methods are based on a novel interpretation and implementation of convex nonparametric least squares (CNLS) which allows not only classification into different clusters but also finding the number of clusters from the data. Moreover, we provide techniques for model validation in CNLS regarding the allocation into groups using efficiency criteria. We provide a prior designed to minimize variation within groups and maximize variation across groups. The new techniques are examined using Monte Carlo experiments and they are applied to a data set of large U.S. banks. Additionally, we propose new techniques for meta-envelopment or meta-frontier formulations in efficiency analysis.

Inverse problems in food engineering: A review

With the increase in the computational power, numerical simulations play an increasing role in designing, development, and optimization of various food processing operations. Inverse and ill-posed problems have been studied extensively in many branches of science and engineering, including mechanical, chemical, aerospace, biology, physics, and chemistry. The inverse techniques have been currently witnessing a growing trend in the food processing field for a decade. Inverse problems are usually performed when direct measurements of heat and mass transfer properties and boundary conditions, are not feasible. They are very sensitive to measurement errors; require optimization methods to tackle the inverse problem. It is necessary to consider the coupling heat and mass transfer for the solution of inverse problem since food process involves simultaneous heat and mass transfer within the food products. To date, inverse methods have been applied in few food processing operations, including drying, baking, freezing, and thermal processing. However, studies related to these areas on the estimation of unknown quantities for various fruits and vegetables, and food products are limited. This review covers the statistical concepts which include confidence interval, confidence region, model validation and minimization techniques discussed in detail. The optimal experimental design, D-optimality and Fisher Information Matrix, and sensitivity analysis are all extensively presented. The optimization techniques and their algorithms used in the area of food processing are explained. Finally, it mainly focuses on the inverse estimation of unknown quantities, namely, heat and mass transfer parameters in different food processing operations.

Data-driven Crowd Modeling Techniques: A Survey

Data-driven crowd modeling has now become a popular and effective approach for generating realistic crowd simulation and has been applied to a range of applications, such as anomaly detection and game design. In the past decades, a number of data-driven crowd modeling techniques have been proposed, providing many options for people to generate virtual crowd simulation. This article provides a comprehensive survey of these state-of-the-art data-driven modeling techniques. We first describe the commonly used datasets for crowd modeling. Then, we categorize and discuss the state-of-the-art data-driven crowd modeling methods. After that, data-driven crowd model validation techniques are discussed. Finally, six promising future research topics of data-driven crowd modeling are discussed.

A Scalable Machine Learning-based Ensemble Approach to Enhance the Prediction Accuracy for Identifying Students at-Risk

Among the educational data mining problems, the early prediction of the students' academic performance is the most important task, so that timely and requisite support may be provided to the needy students. Machine learning techniques may be used as an important tool for predicting low-performers in educational institutions. In the present paper, five single-supervised machine learning techniques have been used, including Decision Tree, Naïve Bayes, k-Nearest-Neighbor, Support Vector Machine, and Logistic Regression. To analyze the effect of an imbalanced dataset, the performance of these algorithms has been checked with and without various resampling methods such as Synthetic Minority Oversampling Technique (SMOTE), Borderline SMOTE, SVM-SMOTE, and Adaptive Synthetic (ADASYN). The Random hold-out method and GridSearchCV were used as model validation techniques and hyper-parameter tuning respectively. The results of the present study indicated that Logistic Regression is the best performing classifier with every balanced dataset generated using all of the four resampling techniques and also achieved the highest accuracy of 94.54% with SMOTE. Furthermore, to improve the prediction results and to make the model scalable, the most suitable classifier was integrated with the help of bagging, and a well-accepted accuracy of 95.45% was achieved.