RESUMEN: Dado que los algoritmos de aprendizaje automático dan resultados más fiables, en los últimos años han sido utilizados en el campo de la salud. Las variables orbitales dan resultados muy exitosos a la hora de clasificar correctamente el sexo. Esta investigación se ha centrado en la determinación del sexo utilizando determinadas variables obtenidas a partir de las imágenes orbitales de la tomografía computarizada (TC) mediante el uso de algoritmos de aprendizaje automático (AA). En este estudio se probaron 12 variables determinadas en 600 imágenes orbitales de 300 individuos (150 hombres y 150 mujeres) con diferentes AA. Se utilizaron algoritmos de AA de árbol de decisión (DT), K-Nearest Neighbour, regresión logística (RL), Random Forest (RF), análisis discriminante lineal (ADL) y Naive Bayes (NB) para el aprendizaje no supervisado. Los análisis estadísticos de las variables se realizaron con el programa Minitab® 21.2 (64 bits). La tasa de ACC de los algoritmos NB, DT, KNN y RL se encontró en % 83, mientras que la tasa de ACC de los algoritmos ADL y RFC se determinó en % 85. Según el análisis de Sharp, la variable con el mayor grado de efecto se encontró como BOW. El estudio determinó el sexo con alta precisión en las proporciones de 0,83 y 0,85 mediante el uso de las variables de las imágenes de TC orbitales, y se adquirieron los datos morfométricos relacionados de la población en cuestión, enfatizando la variación racial.

SUMMARY: Since machine learning algorithms give more reliable results, they have been used in the field of health in recent years. The orbital variables give very successful results in classifying sex correctly. This research has focused on sex determination using certain variables obtained from the orbital images of the computerized tomography (CT) by using machine learning algorithms (ML). In this study 12 variables determined on 600 orbital images of 300 individuals (150 men and 150 women) were tested with different ML. Decision tree (DT), K-Nearest Neighbour (KNN), Logistic Regression (LR), Random Forest (RF), Linear Discriminant Analysis (LDA), and Naive Bayes (NB) algorithms of ML were used for unsupervised learning. Statistical analyses of the variables were conducted with Minitab® 21.2 (64-bit) program. ACC rate of NB, DT, KNN, and LR algorithms was found as % 83 while the ACC rate of LDA and RFC algorithms was determined as % 85. According to Shap analysis, the variable with the highest degree of effect was found as BOW. The study has determined the sex with high accuracy at the ratios of 0.83 and 0.85 through using the variables of the orbital CT images, and the related morphometric data of the population under question was acquired, emphasizing the racial variation.

Resumo O Brasil enfrenta altas taxas de inadimplência, devido em parte à pandemia, levando à busca de novas estratégias de cobranças de dívidas. O Machine Learning (ML), empregado com sucesso em inúmeras áreas, é aliado para elevar a eficácia dessas operações. Este artigo busca apresentar um panorama atual das pesquisas sobre aplicações de ML nas operações de cobrança de dívidas, por meio de uma Revisão Sistemática da Literatura. Foi utilizada a metodologia PICO, identificando inicialmente 41 documentos, dos quais 11 passaram por revisão sistemática. Os resultados mostraram quatro objetivos buscados pelos estudos: previsão de inadimplência, personalização das estratégias de cobrança, otimização das ações de recuperação de dívidas e previsão de recuperação de crédito. E os principais algoritmos utilizados foram Decision Tree, Logistic Regression, Random Forest, Naive Bayes, Artificial Neural Network e Deep Learning. Os resultados revelaram que ML é ainda pouco explorado nessa área, oferecendo potencial para avanços substanciais das pesquisas.

Abstract Brazil is facing high default rates, due in part to the pandemic, leading to the search for new debt collection strategies. Machine Learning (ML), successfully used in numerous areas, is an ally to increase the effectiveness of these operations. This article seeks to present a current overview of research on ML applications in debt collection operations, through a Systematic Literature Review. The PICO methodology was used, initially identifying 41 documents, of which 11 underwent systematic review. The results showed four objectives pursued by the studies: default prediction, personalization of collection strategies, optimization of debt recovery actions and credit recovery prediction. And the main algorithms used were Decision Tree, Logistic Regression, Random Forest, Naive Bayes, Artificial Neural Network and Deep Learning. The results revealed that ML is still little explored in this area, offering potential for substantial research advances.

Resumen: En esta investigación se analizó un conjunto de mensajes de recaudación para el pago de impuestos. En total, se evaluaron doce variaciones de los mensajes de cobro (normas sociales, simplificación, divulgación, compromiso previo, recordatorios y decisiones pasadas) y se determinó su efectividad. Los mensajes fueron transmitidos a microempresarios morosos en cuatro estados brasileños. A partir de una base de datos que contiene información sobre microempresarios morosos, se enviaron 250 000 mensajes de texto para hacer el cobro. Los datos se obtuvieron de la Secretaria Especial da Micro e Pequea Empresa (Sempe). Se utilizaron pruebas para analizar la diferencia entre medias y se utilizó una secuencia de regresión logística. Se utilizaron los widgets Random Forest, regresión logística y Naïve Bayes para definir la solidez del modelo predictivo de Machine Learning. Los resultados de la investigación indicaron que los formatos “simplificación”, “decisiones pasadas” y “alerta” no tuvieron ningún efecto en la lucha contra el incumplimiento. Sin embargo, al combinarlos con mensajes que aludan a normas sociales, los mensajes con el estilo de “decisiones pasadas” y “recordatorios” aumentaron el pago de la deuda. Los widgets empleados mostraron un ajuste excelente al modelo de machine learning. La herramienta Random Forest determinó que el modelo es robusto y adecuado para la función predictiva. Los resultados de la encuesta brindan un aporte a las políticas públicas, pues presentan una acción efectiva para reducir la morosidad tributaria. El uso de mensajes que apelan a las normas sociales puede adaptarse a otras situaciones, y se sugiere aplicarlo en futuras investigaciones.

Abstract: In this research, a group of charging messages for the payment of taxes was investigated. Altogether 12 variations of billing messages (social norms, simplification, disclosure, previous engagement, reminders and previous choices) were evaluated, and their effectiveness was tested. The messages were transmitted to defaulting microentrepreneurs in four Brazilian states. From a database containing information about defaulting micro entrepreneurs, 250 thousand text messages were sent making charges. The data were obtained from the Secretaria Especial da Micro e Pequena Empresa (Sempe). Tests were used to analyse the difference between means and Logistic Regression was used in sequence. The Random Forest, Logistic Regression and Naïve Bayes widgets were used to indicate the robustness of the Machine Learning predictive model. The research findings indicated that the formats “simplification”, “previous choices” and “alert”, employees, did not have an effect in combating default. However, when aligned with social norms, messages in the form of “past options” and “reminders” increase the payment of debts. The widgets used indicated an excellent fit to the machine learning model. The Random Forest tool attested with superiority that the model is robust and suitable for the predictive function. The results of the research provide a contribution to public policies when they present an effective action to reduce defaults in the payment of taxes. The use of messages in the social norm format can be adapted to other situations, constituting a suggestion for future research.

Abstract: Currently, the diversity of sources generating data in a massive online manner cause data streams to become part of many real work applications. Learning from a data stream is a very challenging task due to the non-stationary nature of this type of data. Characteristics such as infinite length, concept drift, concept evolution and recurrent concepts are the most common problems that need to be addressed by data stream learning algorithms. In this work an algorithm for data stream classification based on an associative classifier is presented. This proposal combines a clustering algorithm and the Naïve Associative Classifier for Online Data (NACOD) to address this problem. A set of micro-clusters (MCs), a data structure that summarizes the information of the current data, is used instead of storing the whole data. The MCs are continually updated with the arriving data, either to create new MCs or to update existing ones. The added MCs helps to deal with concept drift. To assess the performance of the proposed model, experiments were carried out on 3 data sets commonly used to evaluate data stream classification algorithms: KDD Cup 1999, Forest Cover Type and Statlog (Shuttle). Our model achieved higher accuracies than those achieved with algorithms such as data stream version of Naïve Bayes and Hoeffding Tree, the average accuracies achieved were for KDD Cup 1999: 100 %, Statlog (Shuttle): 99.01 % and Forest Cover Type 70.44 %.

Abstract: As the world becomes more digitalized, the potential for attacks increases, therefore, effective techniques for intrusion detection on network are needed. In this study, the authors propose a two steps approach. First, the Correlation-based Features Selection as a feature evaluator based on Particle Swarm Optimization is used to select the relevant features. This evaluator is compared with other evaluators. Second, the Random Forest algorithm is used to classify attacks in a network. A comparative study is also performed conducted with different classifiers such as Naïve Bayes, Stochastic Gradient Descent, Deep Learning, k-Nearest Neighbors and Support Vector Machine. Experiments were conducted on the NSL-KDD database and the results show an efficiency of 98.78% for binary classification. The performance results obtained show that the proposed technique performs better than other competing techniques.

Abstract: Tomatoes are widely cultivated, both by family farmers and corporate producers. During the tomato growth cycle, several diseases can affect the plant. The identification of these diseases through short-range images is significant, and computer vision techniques are commonly used to identify diseases in plant leaves. In this paper, a hybrid model that combines a convolutional neural network (CNN) and a Random Forest (RF) decision tree is used for foliar spot detection in tomato leaves. High-level features learned and extracted from CNN are used as input for the RF classifier. To evaluate the proposed model’s performance for plant disease identification, a case study of 2480 low-cost digital RGB images collected in actual field conditions, under different intensities of light exposure, were used, including healthy tomato leaves and leaves with visible symptoms of powdery mildew fungus, which attacks the tomato leaf. The results were compared with six conventional machine learning classifiers: Logistic Regression (LR), Linear Discriminant Analysis (LDA), K- Nearest Neighbors (KNN), Naive Bayes (NB), Support Vector Machine (SVM) and Random Forest (RF). The results show that the proposed model outperformed conventional classifiers, reaching an accuracy of 98%. The results highlight the importance of fusing models to improve the detection plant´s diseases. Abstract cultivated producers cycle shortrange short range significant paper (CNN (RF Highlevel High level classifier s 248 lowcost low cost conditions exposure fungus leaf classifiers LR, LR , (LR) LDA, LDA (LDA) K KNN, KNN (KNN) NB, NB (NB) SVM (SVM RF. . 98 98% plants 24 (LR (LDA (KNN (NB 9 2

Abstract Characterization and development of hydrocarbon reservoirs depends on the classification of lithological patterns from well log data. In thin reservoir units, limited vertical data impedes the efficient classification of lithologies. We present a test case of petrofacies classification using machine learning models in a thin interval of finely laminated limestones using pseudo-well data created over outcrops (radiometric and unconfined compressive strength logs). We tested Gaussian naïve Bayes (GNB) and support vector machine (SVM) techniques to classify eight petrofacies types, divided into two groups. The objective was to observe the capacity of some well-known models to classify petrofacies with a high-frequency vertical variation of diagenetic heterogeneities in an extreme scenario within a thin sedimentary interval. The GNB was less effective (F 1 score of 0.29), and the SVM achieved the best results in classifying the main facies patterns (F 1 = 0.47). However, the GNB performed better when the analysis was focused on distinguishing the two main groups of petrofacies. The results demonstrate that high-frequency facies variations present a challenge to the automatic identification of lithofacies, mainly due to local variations in horizontal heterogeneities (on the mm- to cm-scale) created by depositional and diagenetic processes, which impact the flow in porous media. units lithologies pseudowell pseudo radiometric logs. logs . logs) (GNB (SVM types wellknown known highfrequency high frequency F 0.29, 029 0.29 , 0 29 0.29) 0.47. 047 0.47 47 0.47) However lithofacies mm cmscale cm scale cm-scale processes media 02 0.2 2 04 0.4 4 0.

Abstract: The identification of seeds from native species is a complex assessment due to the high Brazilian biodiversity and varied characteristics between species. The objective was to apply different machine learning classifiers associated with image analysis to identify seeds of forest species. In total, 155 native species belonging to 42 botanical families were analyzed. In addition, to determine the appropriate machine learning classifier, five supervised learning classification techniques were implemented: decision trees (DT), artificial neural networks (ANN), k-nearest neighbors (k-NN), Naive-Bayes classifier (NBC) and support vector machine (SVM), which had their performance evaluated. For modeling, 66% of the seeds’ morphobiometric data were used to train the classifiers, while 34% were reserved for validation. The classifiers are promising tools for identifying species from seed images. The decision tree (DT) classifier showed greater accuracy for correct species identification (82.8%), followed by ANN (81.7%), k-NN (81.7%), NBC (81.1%) and SVM (78.7%). Therefore, it is possible to identify seeds of native species from images and machine learning with a satisfactory accuracy rate. Finally, the decision tree classifier is recommended. Abstract total 15 4 analyzed addition implemented DT, DT , ANN, (ANN) knearest k nearest kNN, kNN NN (k-NN) NaiveBayes Naive Bayes (NBC SVM, (SVM) evaluated modeling 66 34 validation (DT 82.8%, 828 82.8% 82 8 (82.8%) 81.7%, 817 81.7% 81 7 (81.7%) 81.1% 811 1 (81.1% 78.7%. 787 78.7% . 78 (78.7%) Therefore rate Finally recommended (ANN (k-NN (SVM 6 3 82.8 (82.8% 81.7 (81.7% 81.1 (81.1 78.7 (78.7% 82. (82.8 81. (81.7 (81. 78. (78.7 (82. (81 (78. (82 (8 (78 ( (7

Resumo: A identificação de sementes de espécies nativas é uma avaliação complexa devido a elevada biodiversidade brasileira e características variadas entre as espécies. Objetivou-se aplicar diferentes classificadores de aprendizado de máquina associado à análise de imagens para identificar sementes de espécies florestais. Foram analisadas 155 espécies nativas pertencentes a 42 famílias botânicas. Para determinar o classificador de aprendizado de máquina adequado, cinco técnicas de classificação por aprendizado supervisionado foram implementadas: árvores de decisão (DT), redes neurais artificiais (ANN), k-vizinhos mais próximos (k-NN), classificador Naive-Bayes (NBC) e máquina de vetores de suporte (SVM), os quais tiveram seu desempenho avaliados. Para modelagem, 66% dos dados morfobiométricos das sementes foram usados para treinamento dos classificadores, enquanto 34% foram reservados para validação. Os classificadores são ferramentas promissoras para a identificação de espécies a partir das imagens de sementes. O classificador por árvores de decisão (DT) apresentou maior acurácia para identificação correta das espécies (82,8%), seguido dos classificadores ANN (81,7%), k-NN (81,7%), NBC (81,1%) e SVM (78,7%). Portanto, é possível realizar a identificação de sementes de espécies nativas a partir de imagens e aprendizado de máquina com taxa satisfatória de acurácia. Recomenda-se o classificador por árvores de decisão. Resumo Objetivouse Objetivou se florestais 15 4 botânicas adequado implementadas DT, DT , ANN, (ANN) kvizinhos k vizinhos kNN, kNN NN (k-NN) NaiveBayes Naive Bayes (NBC SVM, (SVM) avaliados modelagem 66 34 validação (DT 82,8%, 828 82,8% 82 8 (82,8%) 81,7%, 817 81,7% 81 7 (81,7%) 81,1% 811 1 (81,1% 78,7%. 787 78,7% . 78 (78,7%) Portanto Recomendase Recomenda (ANN (k-NN (SVM 6 3 82,8 (82,8% 81,7 (81,7% 81,1 (81,1 78,7 (78,7% 82, (82,8 81, (81,7 (81, 78, (78,7 (82, (81 (78, (82 (8 (78 ( (7

ABSTRACT: Near-infrared (NIR) spectroscopy is a promising tool for optimizing seed analyses quickly and assertively. The aim of this study was to investigate the viability of NIR in association with chemometric methods in classification of soybean seed lots regarding their physiological potential. We evaluated 372 soybean seed lots for vigor and obtained NIR spectra from seed samples. The original spectra were pre-processed by the following methods: Standard Normal Variate (SNV), SNV + 1st and 2nd derivatives, Gap-segment derivative, and Savitzky-Golay for the first- and second-degree derivatives, as well as combinations of the methods. The lots were divided into Class I (≥ 85% germination after accelerated aging) and Class II (< 85% germination after accelerated aging); and the pre-processed spectra were used to build classification models through the following methods: K-nearest neighbors (KNN), Partial Least Squares - Discriminant Analysis (PLS-DA), Naive Bayes (NB), Random Forest (RF), and Support Vector Machine (SVM). The PLS-DA model showed greater classification accuracy and kappa, followed by SVM. The lowest accuracy values were obtained for the NB and RF models. The regions between the wavelengths 1,000-1,200 nm and 2,200-2,500 nm were the most important for distinguishing the quality levels of soybean seeds. ABSTRACT Nearinfrared Near infrared (NIR assertively potential 37 samples preprocessed pre processed SNV, , (SNV) st nd derivatives Gapsegment Gap segment derivative SavitzkyGolay Savitzky Golay first seconddegree second degree ≥ ( 85 aging < Knearest K nearest KNN, KNN (KNN) PLSDA, PLSDA PLS DA (PLS-DA) NB, (NB) RF, (RF) SVM . (SVM) kappa 1,0001,200 10001200 1,000 1,200 1 000 200 1,000-1,20 2,2002,500 22002500 2,200 2,500 2 500 2,200-2,50 seeds 3 (SNV 8 (KNN (PLS-DA (NB (RF (SVM 0001 1,0001,20 1000120 1000 1,00 1200 1,20 00 20 1,000-1,2 2002 2,2002,50 2200250 2200 2,20 2500 2,50 50 2,200-2,5 1,0001,2 100012 100 1,0 120 1,2 0 1,000-1, 2,2002,5 220025 220 2,2 250 2,5 5 2,200-2, 1,0001, 10001 10 1, 12 1,000-1 2,2002, 22002 22 2, 25 2,200-2 1,0001 1,000- 2,2002 2,200-

RESUMO: A espectroscopia no infravermelho próximo (NIR) consiste em uma ferramenta promissora para otimização das análises de sementes de forma rápida e assertiva. Este trabalho teve como objetivo investigar a viabilidade do NIR, associado a métodos quimiométricos, para classificar lotes de sementes de soja quanto ao potencial fisiológico. Foram utilizados 372 lotes de sementes de soja avaliados quanto ao vigor e obtidos espectros NIR das amostras de sementes. Os espectros originais foram submetidos aos métodos de pré-processamento Standard Normal Variate (SNV), SNV + 1ª e 2ª derivadas; Gap-segment derivative; e Savitzky-Golay, pelas derivadas de primeiro e segundo grau, e a combinação entre os métodos. Os lotes foram divididos em Classe I (≥ 85% de germinação após envelhecimento acelerado), Classe II (< 85% de germinação após envelhecimento acelerado) e os espectros pré-processados foram utilizados para a construção de modelos de classificação por meio dos métodos K-nearest neighbors (KNN), Partial Least Squares - Discriminant Analysis (PLS-DA), Naive Bayes (NB), Random Forest (RF) e Support Vector Machine (SVM). O modelo de classificação PLS-DA apresentou maior acurácia e kappa, seguido pelo SVM. Os menores valores de acurácia foram obtidos para os modelos NB e RF. As regiões entre os comprimentos de ondas 1.000-1.200 nm e 2.200-2.500 nm foram as mais importantes para distinguir os níveis de qualidade das sementes de soja. RESUMO (NIR assertiva quimiométricos fisiológico 37 préprocessamento pré processamento SNV, , (SNV) ª Gapsegment Gap segment derivative SavitzkyGolay, SavitzkyGolay Savitzky Golay, Golay Savitzky-Golay grau ≥ ( 85 acelerado, acelerado < préprocessados processados Knearest K nearest KNN, KNN (KNN) PLSDA, PLSDA PLS DA (PLS-DA) NB, (NB) RF (RF SVM . (SVM) kappa 1.0001.200 10001200 1.000 1.200 1 000 200 1.000-1.20 2.2002.500 22002500 2.200 2.500 2 500 2.200-2.50 3 (SNV 8 (KNN (PLS-DA (NB (SVM 0001 1.0001.20 1000120 1000 1.00 1200 1.20 00 20 1.000-1.2 2002 2.2002.50 2200250 2200 2.20 2500 2.50 50 2.200-2.5 1.0001.2 100012 100 1.0 120 1.2 0 1.000-1. 2.2002.5 220025 220 2.2 250 2.5 5 2.200-2. 1.0001. 10001 10 1. 12 1.000-1 2.2002. 22002 22 2. 25 2.200-2 1.0001 1.000- 2.2002 2.200-

Abstract: The COVID-19 outbreak commenced in Wuhan, China, in December 2019 and swiftly disseminated worldwide. On March 11, 2020, the World Health Organization (WHO) formally designated the COVID-19 outbreak as a global pandemic [1]. This highly contagious disease, which has also started to spread among young people, necessitates the implementation of policies to avert the collapse of hospitals due to shortages in beds, mechanical ventilators, and intensive care units. Using the record from March 2020 to May 31 2021, (before the arrival of vaccines in Mexico) of the website of the General Directorate of Epidemiology of the Ministry of Health of the Government of the Mexican Republic, with more than seven million people associated with COVID-19, a model is built with twelve characteristics that predicts the risk of death associated with COVID-19, applying Supervised Learning Algorithms such as Linear Regression, Naive Bayes, Decision Trees and Random Forests. The results of the machine learning algorithms show a performance of 87%. Subsequently, the model was tested again with 402,116 COVID-19-associated patients from the month of June, achieving an accuracy of 91%, surpassing the model proposed in [3]. When performing data cleaning, we observed that the twelve variables are not correlated, unlike what the authors showed in [4], where data cleaning was not carried out. This analysis can aid in designing strategies and policies to combat its spread and prevent mortality, as well as assist hospitals in prioritizing the care of higher-risk patients.

Abstract: One of the main public health problems is child malnutrition, since it negatively affects the individual throughout his life, limits the development of society and makes it difficult to eradicate poverty. The first objective of this research is to apply data mining techniques for preprocessing, cleaning, reduction and transformation to a data lake that has allowed analyzing anemia in children under 5 years of age, the second objective is to apply Machine Learning algorithms to obtain the best model to predict anemia in children under 5 years of age. The data set was extracted from the open data platform of the government of Peru that corresponds to South Lima, North Lima, East Lima, Central Lima and rural Lima, which collected a total of 138,369 instances and 36 variables of which 30 are categorical and 6 numeric, being an unbalanced data set. In order to obtain the best predictor variables, the Anova F-test and Chi Square filters were used, and it was possible to reduce them to 10 variables, cases were also carried out without considering one of the filters and both filters.To find the best prediction model, the algorithms have been tested: decision tree, logistic regression, K nearest neighbors, random forest and naive bayes. As a result, we show that the best algorithm to predict anemia in children under 5 years of age is the Naive Bayes algorithm with the highest recall of 74%, precision of 43% and accuracy of 70%.

Resumen: Uno de los principales problemas de salud pública es la desnutrición infantil, ya que afecta negativamente al individuo a lo largo de su vida, limita el desarrollo de la sociedad y dificulta la erradicación de la pobreza. El objetivo de esta investigación es aplicar técnicas de minería de datos para el preprocesamiento, limpieza, reducción y transformación a un lago de datos que ha permitido analizar la anemia en niños menores de 5 años, asimismo, se ha aplicado algoritmos de Machine Learning para obtener el mejor modelo que permita predecir la anemia en niños menores de 5 años. El conjunto de datos fue extraído de la plataforma datos abiertos del gobierno del Perú que corresponde a Lima sur, Lima Norte, Lima Este, Lima Centro y Lima rural que se juntaron en total 138369 instancias y 36 variables, de la cuales 30 son categóricas y 6 numéricas, siendo un conjunto de datos desequilibrado. Para poder obtener las mejores variables predictoras se utilizó los filtros Anova F-test y Chi Cuadrado y se logró reducir a 10 variables, también se realizó casos sin considerar uno de los filtros y ambos filtros. Para buscar el mejor modelo de predicción se ha probado los algorítmicos: árbol de decisión, regresión logística, K vecinos más cercanos, bosque aleatorio y Naive bayes. Como resultado mostramos que el mejor algoritmo que permite predecir la anemia en niños menores de 5 años es el Naive Bayes con mayor recall de 74 %, precisión de 43% y exactitud del 70 %.

ABSTRACT The ranking of seed lots is a fundamental process for all companies in the seed industry. This work aims to demonstrate data mining methods for ranking sorghum seed lots during the seed processing through analysis of quality control data. Germination and cold tests were performed to verify the physiological quality of the lots. Seed samples from each lot were evaluated in two moments: post-cleaning and finished product (ready for marketing). The results after pre-processing totaled 188 rows of data with six attributes, encompassing 150 lots accepted for marketing, 6 rejected, and 32 intermediate lots. The classifiers used were J48, Random Forest, Classification Via Regression, Naive Bayes, Multilayer Perceptron, and IBk. The Resample filter was used for adjustment of the data. The k-fold technique was used for training, with ten folds. The metrics of Accuracy, Precision, Recall, F-measure, and ROC Area were used to verify the accuracy of the algorithms. The results obtained were used to determine the best machine-learning algorithm. IBk and J48 presented the highest accuracy of data; the IBk technique presented the best results. The Resample filter was essential for solving the data imbalance problem. Sorghum seed lots can be classified with great accuracy and precision through artificial intelligence and machine learning technique. industry moments postcleaning post cleaning ready marketing. marketing . marketing) preprocessing pre 18 attributes 15 rejected 3 J Forest Regression Bayes Perceptron kfold k fold training folds Accuracy Precision Recall Fmeasure, Fmeasure F measure, measure F-measure algorithms machinelearning algorithm J4 problem 1

RESUMO A classificação de lotes de sementes é um processo fundamental para todas as empresas do setor sementeiro. O objetivo do trabalho é demonstrar os métodos de mineração de dados de ranqueamento de lotes de sementes de sorgo durante o processo de beneficiamento, através de análises de dados do controle de qualidade. Os testes realizados foram germinação e teste de frio, com o objetivo de verificar a qualidade fisiológica dos lotes. As amostras de sementes de cada lote foram avaliadas em dois momentos: póslimpeza e produto acabado (pronto para comercialização). Os dados gerados, após o pré-processamento, totalizaram 188 linhas com seis atributos, contabilizando 150 lotes aceitos para comercialização, seis rejeitados e 32 denominados intermediários. Os classificadores utilizados foram J48, Random Forest, Classification Via Regression, Naive Bayes, Multilayer Perceptron e IBk. Utilizou-se o filtro Resample para ajustamento dos dados. A técnica empregada para treinamento foi a k-fold, com 10 folds. Para verificar a precisão dos algoritmos foram utilizadas as métricas de Acurácia, Precisão, Recall, F-measure e Área ROC. Com os resultados obtidos determinou-se o melhor algoritmo de aprendizagem de máquina. Verificou-se que o IBk e o J48 obtiveram maior acurácia nos dados, sendo que a técnica de IBk obteve o melhor resultado. O filtro Resample foi importante para resolver o problema do desequilíbrio dos dados. Concluímos ser possível classificar lotes de sementes de sorgo com grande acurácia e precisão através de inteligência artificial e sua técnica de aprendizado de máquina. sementeiro beneficiamento frio momentos pronto comercialização. comercialização . comercialização) gerados préprocessamento, préprocessamento pré processamento, processamento pré-processamento 18 atributos 15 3 intermediários J Forest Regression Bayes Utilizouse Utilizou se kfold, kfold k fold, fold k-fold 1 folds Acurácia Precisão Recall Fmeasure F measure ROC determinouse determinou máquina Verificouse Verificou J4 resultado

Abstract The novel coronavirus disease (COVID-19) is an ongoing pandemic with large global attention. However, spreading fake news on social media sites like Twitter is creating unnecessary anxiety and panic among people towards this disease. In this paper, we applied machine learning (ML) techniques to predict the sentiment of the people using social media such as Twitter during the COVID-19 peak in April 2021. The data contains tweets collected on the dates between 16 April 2021 and 26 April 2021 where the text of the tweets has been labelled by training the models with an already labelled dataset of corona virus tweets as positive, negative, and neutral. Sentiment analysis was conducted by a deep learning model known as Bidirectional Encoder Representations from Transformers (BERT) and various ML models for text analysis and performance which were then compared among each other. ML models used were Naïve Bayes, Logistic Regression, Random Forest, Support Vector Machines, Stochastic Gradient Descent and Extreme Gradient Boosting. Accuracy for every sentiment was separately calculated. The classification accuracies of all the ML models produced were 66.4%, 77.7%, 74.5%, 74.7%, 78.6%, and 75.5%, respectively and BERT model produced 84.2%. Each sentiment-classified model has accuracy around or above 75%, which is a quite significant value in text mining algorithms. We could infer that most people tweeting are taking positive and neutral approaches.

Resumen En este artículo, aplicamos técnicas de aprendizaje automático para predecir el sentimiento de las personas que usan las redes sociales como Twitter durante el pico de COVID-19 en abril de 2021. Los datos contienen tweets recopilados en las fechas entre el 16 de abril de 2021 y el 26 de abril de 2021, donde el texto de los tweets se ha etiquetado mediante la formación de los modelos con un conjunto de datos ya etiquetado de tweets de virus de corona como positivo, negativo y neutro. El análisis del sentimiento se llevó a cabo mediante un modelo de aprendizaje profundo conocido como Representaciones de Codificadores Bidireccionales de Transformers (BERT) y varios modelos de aprendizaje automático para el análisis de texto y el rendimiento, que luego se compararon entre sí. Los modelos ML utilizados son Bayes ingenuas, regresión logística, bosque aleatorio, máquinas vectoriales de soporte, descenso de gradiente estocástico y aumento de gradiente extremo. La precisión de cada sentimiento se calculó por separado. La precisión de clasificación de todos los modelos de ML producidos fue de 66.4 %, 77.7 %, 74.5 %, 74.7 %, 78.6 % y 75.5 %, respectivamente y el modelo BERT produjo 84.2%. Cada modelo clasificado de sentimiento tiene una precisión de alrededor o superior al 75 %, que es un valor bastante significativo en los algoritmos de minería de texto. Vemos que la mayoría de las personas que tuitean están adoptando un enfoque positivo y neutral.

Abstract Introduction: to predict work related diseases is a challenge for organizations and the governmental authorities. By means of machine learning (ML) techniques it is possible to identify factors that determine the occurrence of an occupational disease, aiming at taking more effective actions to protect workers. Objective: to predict, by comparing ML techniques, the factors which highly influence the occurrence of occupational dermatitis. Methods: we developed a code in R language and a descriptive analysis of the data and identified the influence factors according to the ML technique that presented the best performance. The database was made available by the Occupational Dermatology Service of Oswaldo Cruz Foundation and assembles information of the workers who experienced cutaneous alterations suggestive of occupational dermatitis between 2000-2014. Results: the techniques which presented the best performance were: neural network, random forest, support vector machine, and naive Bayes. Sex, schooling, and profession were the most adequate variables for the occupational dermatitis prediction models. Conclusion: ML techniques allowed to predict the factors that influence the workers’ safety and health, as well as the parameters that subsidize the procedures implementation, and the most effective policies to prevent occupational dermatitis. Introduction authorities (ML disease Objective Methods 20002014. 20002014 2000 2014. 2014 2000-2014 Results network forest Bayes Sex schooling models Conclusion health implementation 2000201 200 201 2000-201 200020 20 2000-20 20002 2 2000-2 2000-

Resumo Introdução: realizar a predição de doenças relacionadas ao trabalho é um desafio às organizações e ao poder público. Com as técnicas de aprendizado de máquina (AM), é possível identificar fatores determinantes para a ocorrência de uma doença ocupacional, visando direcionar ações mais efetivas à proteção dos trabalhadores. Objetivo: predizer, a partir da comparação de técnicas de AM, os fatores com maior influência para a ocorrência de dermatite ocupacional. Métodos: desenvolveu-se um código em linguagem R e uma análise descritiva dos dados e identificaram-se os fatores de influência de acordo com a técnica de AM que demonstrou melhor desempenho. O banco de dados foi disponibilizado pelo Serviço de Dermatologia Ocupacional da Fundação Oswaldo Cruz e contém informações de trabalhadores que apresentaram alterações cutâneas sugestivas de dermatite ocupacional no período de 2000-2014. Resultados: as técnicas com melhor desempenho foram: neural network, random forest, support vector machine e naive Bayes. As variáveis sexo, escolaridade e profissão foram as mais adequadas para os modelos de previsão de dermatite ocupacional. Conclusão: as técnicas de AM possibilitam predizer os fatores que influenciam a segurança e a saúde dos trabalhadores, os parâmetros que subsidiam a implantação de procedimentos e as políticas mais efetivas para prevenir a dermatite ocupacional. Introdução público , (AM) Objetivo Métodos desenvolveuse desenvolveu se identificaramse identificaram 20002014. 20002014 2000 2014. 2014 2000-2014 Resultados network forest Bayes sexo Conclusão (AM 2000201 200 201 2000-201 200020 20 2000-20 20002 2 2000-2 2000-

Resumen Objetivo Comparar el desempeño de modelos de aprendizaje de máquina con Medication Fall Risk Score (MFRS) para la previsión del riesgo de caída relacionado con medicamentos prescriptos. Métodos Se trata de un estudio caso-control retrospectivo de pacientes adultos y adultos mayores de un hospital terciario de Porto Alegre, estado de Rio Grande do Sul, Brasil. Se investigaron los medicamentos prescriptos y las clases de medicamentos. Los datos fueron exportados al software RStudio para el análisis estadístico. Las variables se analizaron a través de los algoritmos de regresión logística Naive Bayes, Random Forest y Gradient Boosting. La validación del algoritmo se realizó usando validación cruzada de 10 veces. El índice de Youden fue la métrica seleccionada para evaluar los modelos. El proyecto fue aprobado por el Comité de Ética en Investigación. Resultados El modelo de aprendizaje de máquina que presentó el mejor desempeño fue el desarrollado por el algoritmo Naive Bayes. El modelo construido a partir de un conjunto de datos de un hospital específico presentó mejores resultados en la población estudiada que el MFRS, una herramienta generalizada. Conclusión Herramientas de previsión de riesgo que dependen de la aplicación y el registro adecuados por parte de los profesionales demandan tiempo y atención que podría ser destinado al cuidado del paciente. Modelos de previsión construidos mediante algoritmos de aprendizaje de máquina pueden ayudar a identificar riesgos para mejorar la atención al paciente. MFRS (MFRS casocontrol caso control Alegre Sul Brasil estadístico Bayes Boosting 1 veces Investigación generalizada paciente

Abstract Objective To compare the performance of machine-learning models with the Medication Fall Risk Score (MFRS) in predicting fall risk related to prescription medications. Methods This is a retrospective case-control study of adult and older adult patients in a tertiary hospital in Porto Alegre, RS, Brazil. Prescription drugs and drug classes were investigated. Data were exported to the RStudio software for statistical analysis. The variables were analyzed using Logistic Regression, Naive Bayes, Random Forest, and Gradient Boosting algorithms. Algorithm validation was performed using 10-fold cross validation. The Youden index was the metric selected to evaluate the models. The project was approved by the Research Ethics Committee. Results The machine-learning model showing the best performance was the one developed by the Naive Bayes algorithm. The model built from a data set of a specific hospital showed better results for the studied population than did MFRS, a generalizable tool. Conclusion Risk-prediction tools that depend on proper application and registration by professionals require time and attention that could be allocated to patient care. Prediction models built through machine-learning algorithms can help identify risks to improve patient care. machinelearning machine learning MFRS (MFRS medications casecontrol case control Alegre RS Brazil investigated analysis Regression Forest 10fold fold 10 Committee algorithm tool Riskprediction prediction care 1

Resumo Objetivo Comparar o desempenho de modelos de aprendizado de máquina com o Medication Fall Risk Score (MFRS) na previsão de risco de queda relacionado a medicamentos prescritos. Métodos Trata-se de um estudo caso-controle retrospectivo de pacientes adultos e idosos de um hospital terciário de Porto Alegre, RS, Brasil. Medicamentos prescritos e classes de medicamentos foram investigados. Os dados foram exportados para o software RStudio para análise estatística. As variáveis foram analisadas por meio dos algoritmos de Regressão Logística, Naive Bayes, Random Forest e Gradient Boosting. A validação do algoritmo foi realizada usando validação cruzada de 10 vezes. O índice de Youden foi a métrica selecionada para avaliar os modelos. O projeto foi aprovado pelo Comitê de Ética em Pesquisa. Resultados O modelo de aprendizado de máquina que apresentou melhor desempenho foi o desenvolvido pelo algoritmo Naive Bayes. O modelo construído a partir de um conjunto de dados de um hospital específico apresentou melhores resultados para a população estudada do que o MFRS, uma ferramenta generalizável. Conclusão Ferramentas de previsão de risco que dependem de aplicação e registro adequados por parte dos profissionais demandam tempo e atenção que poderiam ser alocados ao cuidado do paciente. Modelos de previsão construídos por meio de algoritmos de aprendizado de máquina podem ajudar a identificar riscos para melhorar o atendimento ao paciente. MFRS (MFRS Tratase Trata se casocontrole caso controle Alegre RS Brasil investigados estatística Logística Bayes Boosting 1 vezes Pesquisa generalizável paciente