Navegando por Data de Publicação, começando com "2021-10-22"

Filtrar resultados por ano ou mês

Agora exibindo 1 - 5 de 5

Método bagging para aprimoramento de previsões de séries temporais
(Universidade Federal de São Carlos, 2021-10-22) Camargo, Juliana Shibaki; Andrade Filho, Marinho Gomes de; https://lattes.cnpq.br/4126245980112687; Diniz, Carlos Alberto Ribeiro; https://lattes.cnpq.br/3277371897783194; https://lattes.cnpq.br/1432636060298365
Different methodologies are proposed and explored aiming to reduce time series forecasting error. A promising approach consists in combining different forecasts from different models in order to get a better accuracy, i.e., a smaller forecast error. This work aims to review and apply the bootstrap aggregating method, also known as bagging, in order to improve time series forecasting. First, each time series is divided into training and testing time series, and then the moving block bootstrap methodology is applied to the training series to generate different resampled time series, and then forecasting for each one of the series is performed and combined, thus obtaining the final combined forecast. The test data set is used to calculate the accuracy of the models, individual and combined. A simulation study of time series and application to a real time series data sets are presented. The chosen and fitted model for each of the time series was an autoregressive integrated moving average (ARIMA). The accuracy measurements used were the mean square error and its root, mean arctangent absolute percentage error and the symmetric mean absolute percentage error. Finally, the impact on the forecasts of the combined model by varying the resampling method parameters was explored and comparisons between the combined and individual forecasting methods were also carried out.
Desenvolvimento e aplicação de módulo de sensoriamento de baixo custo para o monitoramento de saúde estrutural de máquina rotativa
(Universidade Federal de São Carlos, 2021-10-22) Mascoloti, Gabriela Tavares; Shiki, Sidney Bruce; https://lattes.cnpq.br/0573973677787523; https://lattes.cnpq.br/8487846064391835
Predictive maintenance is a recent maintenance model on the market, although it is already well known in the aeronautical industry, and it is increasingly necessary in any manufacturing environment. The research project aims to develop low-cost reduction nodes for predictive monitoring of machine health, increasing production performance and reducing long-term maintenance costs. Integrated with the concept of Internet of Things (IoT) and data processing, it will result in a tool to increase machine reliability. Precisely because it is a low-cost development, which will be developed in a laboratory with hardware and commercial sensor, the main challenges of the project is the dimensioning in such a way that it meets the requirements of an industrial environment and is economically advantageous. In addition, once developed, the technical staff responsible for monitoring must be trained to correctly read the information that will be treated and provided from a communication platform with the cloud. Another challenge of the research is to be able to install the prototype developed in a factory and prove the efficiency of the method. The focus of the main research is to develop a predictive monitoring device that is easy to install and operate with low cost, integrated with data treatment via the cloud and possible Application for industry. Once this goal is achieved, the device will be tested in a laboratory and industrial environment for study validation and data comparison. The study will focus on comparing the performance of machines with and without monitoring via a device developed in the research, in order to prove the importance and necessity of predictive maintenance aligned with data processing in the industry.
A robust lasso regression for linear mixed-effects models with diagnostic analysis
(Universidade Federal de São Carlos, 2021-10-22) Garcia, Rafael Rocha de Oliveira; Novelli, Cibele Maria Russo; https://lattes.cnpq.br/1011098065426388; https://lattes.cnpq.br/0412360055686637
Variable selection has been a topic of great interest for statisticians and researchers alike. The choice of the best subset of predictors may be carried out with the objective of improving prediction or for easier interpretation of results. However, such methods are not always straightforward, mainly in the context of linear mixed-effects models. Variable selection for such models must be carried out for both fixed and random effects, the first being related to the global mean of data and the second to subject-level variance. There are two possible approaches when selecting variables for mixed-effects models: joint or two-stage procedures. In existing literature on the topic of variable selection for linear mixed-effects model, there is a method of joint selection via lasso for linear mixed-effects models under a normal distribution. Another topic of remarkable importance, is diagnostics and residual analysis. While residual analyses are carried out to assess issues with the fitted model and identification of atypical observations, diagnostic analyses are carried out assuming the model as correct and, assessing its conclusions robustness to small disturbances in the data and/or the model. There are many possible ways to deal with such observations. One is using robust models, which are said to be robust to disturbances in the data. That is, models that are better fit to data sets that possess observations considered to be as outliers and/or leverage. This work aims to use the robust method for variable selection in linear mixed-effects model and compare it with the normal method using diagnostic analysis.
Produção biotecnológica de galacto-oligossacarídeos (GOS) a partir do soro do queijo porungo
(Universidade Federal de São Carlos, 2021-10-22) Bolognesi, Lais Saldanha; Gabardo, Sabrina; https://lattes.cnpq.br/9732593816895465; https://lattes.cnpq.br/5678265156332430
The use of agro-industrial residues associated with enzymatic process technology has encouraged the development of new strategies for obtaining bioproducts, such as galactooligosaccharides (GOS). The β-galactosidase enzyme is widely used on production of lactose-free products and, recently, studies on the production of GOS have emerged to take advantage of the main residue of dairy industries – cheese whey. This is a substrate rich in lactose, contributing as an alternative and low-cost source for obtaining GOS, oligosaccharides that act as prebiotics. One way to obtain GOS is using the enzymatic immobilization technique, which allows for several advantages, including the reuse of the biocatalyst, the reduction of maintenance costs and the improvement of thermal stability. In this context, this project aimed to study the use of porungo cheese whey, a dairy product characteristic of the southwest of São Paulo, to produce GOS through the immobilized β-galactosidase enzyme. The work was divided into two moments, in which, in the first, two enzymatic immobilization strategies were evaluated (calcium alginate and calcium alginate-Concanavalin A) in relation to the optimum pH and temperature of the enzyme. In the second moment, obtaining GOS was tested using the most promising immobilization support on different substrates (lactose solution and cheese whey) and then from different concentrations of lactose added to porungo cheese whey (200 g L-1 to 400 g L-1) and temperatures (37 °C to 55 °C). Enzyme immobilization allowed us to verify an increase in the optimal pH range (7.0 to 7.5) for the calcium alginate support when compared to the free enzyme pH range (6.5), while for the calcium alginate support Calcium-Concanavalin A, a greater enzymatic activity was observed at lower pHs (6.0). With respect to temperature, a greater enzymatic activity was observed for the free enzyme in relation to the two immobilization supports tested. Higher GOS yields were observed in porungo cheese whey (63.2%) compared to the lactose solution (41.1%), which was used in the next step to evaluate the process conditions. The maximum GOS yield was 17.73%, with 61.4% lactose conversion, at 46 °C and 400 g L-1. These results suggest the possible use of porungo cheese whey as a substrate in the biotechnological production of GOS.
Formação e manutenção de classes de equivalência: efeitos da magnitude das consequências e do uso de um procedimento gamificado
(Universidade Federal de São Carlos, 2021-10-22) Santos, Alceu Regaço dos; Souza, Deisy das Graças de; https://lattes.cnpq.br/4404800720856419; https://lattes.cnpq.br/9728855417947699
Studies have showed that the relations between equivalence class stimuli can vary in function of numerous experimental parameters. However, few studies investigated the effect of consequences as an independent variable. This study is divided in three experiments, which have as general purpose to evaluate the effect of using reinforcing and punishing consequences in the formation and maintenance of equivalence relations. The Experiment 1 and 2 compared the use of contingencies of reinforcement and punishment exclusively or together, for training two potential classes of stimuli with four member each. It was used a traditional Matching-to-Sample procedure and the participants were distributed in three groups related with the consequences used: Reinforcement Group (R), Reinforcement-Punishment Group (R-P) and Punishment Group (G). A bigger number of participants in Group R reached the criteria for the formation of the equivalence classes when compared with the other groups. Experiment 3 had as specific objective an evaluation of the effect of the magnitude of the reinforcement and punishment consequences in the formation and maintenance of equivalence classes. For data collection, a gamified version of the MTS (the game Miner Troubles, developed for the study) was used, adapted for online data collection, due to limitations resulting from the Covid-19 pandemic. 56 people (between 18 and 57 years old) participated in the research, randomly distributed among three groups: the More-Earning Group (M-E), the Balanced Group (B) and the More-Loss Group (M-L). In the first group, the participants gained 4 points when they choose correctly and lost 1 point when making a mistake; in the second group, the participants gained 1 point when they choose correctly and lost 1 point when making a mistake; and in the third group, the participants gained 1 point when they choose correctly and lost 4 points when they choose incorrectly. As the results, the use of a greater magnitude of punishment interfered in the learning of conditional relationships and in the emergence of untrained relationships: the performance of M-L participants was less accurate in the Symmetry and Relationship Maintenance Tests compared to other groups. This experiment extends the findings of other studies on the effect of punishment on discriminative learning and gives rise to a discussion about the use of punishment more intensely in learning.