Efficient multidisciplinary modeling of aircraft undercarriage landing gear using data-driven Naïve Bayes and finite element analysis

Multiscale and Multidisciplinary Modeling Experiments and Design, Journal Year: 2024, Volume and Issue: 7(4), P. 3187 - 3199

Published: March 16, 2024

Language: Английский

---

Predicting dairy cattle heat stress using machine learning techniques

Journal of Dairy Science, Journal Year: 2020, Volume and Issue: 104(1), P. 501 - 524

Published: Oct. 31, 2020

Language: Английский

---

Predicting Rice Pest Population Occurrence with Satellite-Derived Crop Phenology, Ground Meteorological Observation, and Machine Learning: A Case Study for the Central Plain of Thailand

Applied Sciences, Journal Year: 2019, Volume and Issue: 9(22), P. 4846 - 4846

Published: Nov. 12, 2019

The brown planthopper Nilaparvata lugens (BPH) is one of the most harmful insect pests in rice paddy fields, which causes considerable yield loss and consequent economic problems, particularly central plain Thailand. Accurate timely forecasting pest population incidence would support farmers planning effective mitigation. In this study, artificial neural network (ANN), random forest (RF) classic linear multiple regression (MLR) analyses were applied compared to forecast BPH using weather host-plant phenology factors during crop dry season from 2006 2016 Data satellite earth observation was used monitor affecting density. An ANN model with integrated ground-based meteorological variables satellite-derived host plant more accurate for short-term peak abundance when RF MLR, according a reasonably validating dataset (RMSE natural log-transformed (ln) light trap catches = 1.686, 1.737, 2.015, respectively). This finding indicates that utilization ground observations, NDVI time series, have potential predict density management programs. We expect results study can be conjunction satellite-based monitoring system developed by Geo-Informatic Space Technology Development Agency Thailand (GISTDA) an early warning system.

Language: Английский

---

Living With COVID-19: A Systemic and Multi-Criteria Approach to Enact Evidence-Based Health Policy

Frontiers in Public Health, Journal Year: 2020, Volume and Issue: 8

Published: June 16, 2020

The lifting of COVID-19 (coronavirus disease 2019) lockdown requires, in the short and medium terms, a holistic evidence-based approach to population health management based on combining risk factors bio-economic outcomes, including actors' behaviors. This dynamic global control is necessary deal with new paradigm living an infectious disease, which disrupts our individual freedom challenge for policymakers consists defining methods lockdown-lifting follow-up (middle-term rules) that best meet needs resumption economic activity, societal wellbeing, containment outbreak. There no simple ready-to-use way do this since it means considering several competing objectives at same time continuously adapting strategy rules, ideally local scale. We propose framework creating precision policy simultaneously considers public health, economic, dimensions while accounting constraints uncertainty. It four following principles: integrating multiple heterogeneous information, accepting navigation uncertainty, adjusting dynamically feedback mechanisms, managing clusters through multi-scalar conception. intervention obtained includes scientific background via epidemiological modeling modeling. A set quantitative qualitative indicators are used as precisely monitor societal-economic-epidemiological dynamics, allowing tightening or loosening measures before epidemic damage (re-)occurs. Altogether, allows steers avoids any political shock.

Language: Английский

---

A Comparison of Naive Bayes Methods, Logistic Regression and KNN for Predicting Healing of Covid-19 Patients in Indonesia

Published: April 9, 2021

Since it was declared a global pandemic by the World Health Organization (WHO), number of cases Covid-19 patients who died has continued to increase. One countries with highest death rate in world is Indonesia. On Saturday, April 4, 2020, Indonesia reached for patients, around 9.11%. This must be suppressed so that there are no more victims. For this reason, necessary know actually factors can reduce risk and predict chance curing patients. In data mining, several methods used patient's recovery considering variables. The variables study were age gender. Naive Bayes Method, logistic regression, K-Nearest Neighbor (KNN) chosen analyze their most accurate performance. result shows KNN accuracy, which 0.750 compared regression value 0.703 as well same value. Meanwhile, level precision three models also value, namely than have 0.700. recall kNN remains two comparison 0.708.

Language: Английский

---

Over 20 Years of Machine Learning Applications on Dairy Farms: A Comprehensive Mapping Study

Sensors, Journal Year: 2021, Volume and Issue: 22(1), P. 52 - 52

Published: Dec. 22, 2021

Machine learning applications are becoming more ubiquitous in dairy farming decision support areas such as feeding, animal husbandry, healthcare, behavior, milking and resource management. Thus, the objective of this mapping study was to collate assess studies published journals conference proceedings between 1999 2021, which applied machine algorithms farming-related problems identify trends geographical origins data, well algorithms, features evaluation metrics methods used. This carried out line with PRISMA guidelines, six pre-defined research questions (RQ) a broad unbiased search strategy that explored five databases. In total, 129 publications passed selection criteria, from relevant data required answer each RQ were extracted analyzed. found Europe (43% studies) produced largest number (RQ1), while articles Computers Electronics Agriculture journal (21%) (RQ2). The addressed related physiology health cows (32%) (RQ3), most frequently employed feature derived sensors (48%) (RQ4). tree-based (54%) (RQ5), RMSE (56%) (regression) accuracy (77%) (classification) used, hold-out cross-validation (39%) method (RQ6). Since 2018, there has been than sevenfold increase focused on cows, compared almost threefold overall publications, suggesting an increased focus subdomain. addition, fivefold neural network identified since comparison use both statistical regression increasing utilization network-based algorithms.

Language: Английский

---

Effectiveness of machine learning algorithms as a tool to meat traceability system. A case study to classify Spanish Mediterranean lamb carcasses

Food Control, Journal Year: 2024, Volume and Issue: 164, P. 110604 - 110604

Published: May 29, 2024

Establishing the traceability of meat products has been a major focus food science in recent decades. In this context, advances nutritional biomarker identification and improvements statistical technology have allowed for more accurate classification products. Moreover, artificial intelligence now provided new opportunity optimizing existing methods to identify animal This study presents comparative analysis effectiveness different machine learning algorithms based on raw data from analyses organoleptic, sensory traits differentiate categories commercial lamb an indigenous Spanish breed (Mallorquina breed) obtained following production systems: suckling lambs; light lambs grazing; grazing supplemented with grain. Six were evaluated: Artificial Neural Network (ANN), Decision Tree, K-Nearest Neighbours (KNN), Naive Bayes, Multinomial Logistic Regression, Support Vector Machine (SVM). For each algorithm, we tested three datasets, namely organoleptic sensorial (CIELAB colour, water holding capacity, Warner-Bratzler shear force, volatile compounds trained tasters), (proximate composition fatty acid profile). We also combination all datasets. All combined into dataset 144 variables resulting characterization, which included 11,232 event records. The ANN algorithm stood out its high score datasets used. fact, overall accuracy 0.88, 0.83, 0.88 organoleptic-sensory, nutritional, respectively. using SVM assign according system performed better full performances equal those ANN. KNN showed worst performance, accuracies 0.54 or lower results demonstrate that is useful tool classifying carcasses. could be proposed as tools differentiating characteristics Mediterranean lambs' meat. However, order improve systems guarantee consumers processes used by these algorithms, studies along lines other breeds are required.

Language: Английский

---

Can machine learning algorithms perform better than multiple linear regression in predicting nitrogen excretion from lactating dairy cows

Scientific Reports, Journal Year: 2022, Volume and Issue: 12(1)

Published: July 21, 2022

This study aims to compare the performance of multiple linear regression and machine learning algorithms for predicting manure nitrogen excretion in lactating dairy cows, develop new prediction models MN excretion. Dataset used were collated from 43 total diet digestibility studies with 951 cows. Prediction developed evaluated using MLR technique three algorithms, artificial neural networks, random forest support vector regression. The ANN model produced a lower RMSE higher CCC, compared MLR, RFR SVR model, tenfold cross validation. Meanwhile, hybrid knowledge-based data-driven approach was implemented selecting features this study. Results showed that greatly improved by turning process selection algorithms. proposed intake as primary predictor. Alternative also based on live weight milk yield use condition where data are not available (e.g., some commercial farms). These provide benchmark information mitigation under typical production conditions managed within grassland-based systems.

Language: Английский

---

Machine learning outperformed logistic regression classification even with limit sample size: A model to predict pediatric HIV mortality and clinical progression to AIDS

PLoS ONE, Journal Year: 2022, Volume and Issue: 17(10), P. e0276116 - e0276116

Published: Oct. 14, 2022

Logistic regression (LR) is the most common prediction model in medicine. In recent years, supervised machine learning (ML) methods have gained popularity. However, there are many concerns about ML utility for small sample sizes. this study, we aim to compare performance of 7 algorithms 1-year mortality and clinical progression AIDS a cohort infants living with HIV from South Africa Mozambique. The data set (n = 100) was randomly split into 70% training 30% validation set. Seven (LR, Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naïve Bayes (NB), Artificial Neural Network (ANN), Elastic Net) were compared. variables included as predictors same across models including sociodemographic, virologic, immunologic, maternal status features. For each models, parameter tuning performed select best-performing hyperparameters using 5 times repeated 10-fold cross-validation. A confusion-matrix built assess their accuracy, sensitivity, specificity. RF ranked best algorithm terms accuracy (82,8%), sensitivity (78%), AUC (0,73). Regarding specificity showed better than other external highest AUC. LR lower compared RF, SVM, or KNN. outcome children perinatally acquired can be predicted considerable algorithms. Better would benefit less specialized staff limited resources countries improve prompt referral case high-risk progression.

Language: Английский

---

Machine Learning to Predict Pregnancy in Dairy Cows: An Approach Integrating Automated Activity Monitoring and On-Farm Data

Animals, Journal Year: 2024, Volume and Issue: 14(11), P. 1567 - 1567

Published: May 25, 2024

Automated activity monitoring (AAM) systems are critical in the dairy industry for detecting estrus and optimizing timing of artificial insemination (AI), thus enhancing pregnancy success rates cows. This study developed a predictive model to improve by integrating AAM data with cow-specific environmental factors. Utilizing from 1,054 cows, this compared outcomes between two AI timings—8 or 10 h post-AAM alarm. Variables such as age, parity, body condition, locomotion, vaginal discharge scores, peripartum diseases, breeding program, bull used AI, milk production at time conditions (season, relative humidity, temperature–humidity index) were considered alongside on rumination, activity, intensity. Six models assessed determine their efficacy predicting success: logistic regression, Bagged AdaBoost algorithm, linear discriminant, random forest, support vector machine, Classification Tree. Integrating on-farm significantly enhanced prediction accuracy using alone. The forest showed superior performance, highest Kappa statistic lowest false positive rates. discriminant regression demonstrated best accuracy, minimal negatives, area under curve. These findings suggest that combining can reproductive management industry.

Language: Английский

---