Principal Component Analysis PCA Research Articles

Measuring A Composite Financial Inclusion Index in Lower-Middle Income Countries

This study suggests a composite index of financial inclusion (FI) for financial development, based on three key dimensions: penetration, availability, and usage in lower-middle income countries. The methodology uses principal component analysis PCA to exploit a composite index of financial inclusion and test its impact of new components on financial development. On the secondary data of financial access from the sources of FAS-IMF and WB-Findex, the research result is found that there are two new components of overall financial inclusion: the traditional financial inclusive index (TFI) includes three traditional dimensions: traditional penetration (TP), traditional availability (TA), and traditional usage (TU); and the digital financial inclusive index (DFI) includes three digital dimensions: digital penetration (DP), digital availability (DA), and digital usage (DU). The research findings give the empirical evidence that the digitalized level of financial inclusion is quite weak in lower-middle income countries. The research contributes to the methodology of building a composite index of financial inclusion with optimal weights of dimensions and traditional-digital financial inclusive indexes.

Discrimination model of geographical area from coconut milk by near-infrared spectroscopy: Exploration in tandem with classical chemometrics, machine learning, and deep learning

This work proposes exploring the discrimination model by near-infrared (NIR) spectroscopy (FT-NIR and Micro-NIR) for geographical source areas of coconut milk in tandem with the classical to modern chemometrics classifier. The discrimination model was developed using qualitative chemometrics techniques from classic (Principal Component Analysis-PCA, Partial Least Squares Discriminant Analysis-PLS-DA, Linear Discriminant Analysis-LDA) to modern, including classifiers from machine learning (Support Vector Machine-SVM, k-Nearest Neighbor-KNN, Artificial Neural Network-ANN) and deep learning (Simple Convolutional Neural Networks-S-CNN, S-AlexNET, Residual Networks-ResNET). Three sources as geographical areas of coconut milk originally from Thailand were used, including the south region (Chumphon Province), middle region (Samut Songkhram Province), and east region (Chonburi Province). Our findings showed that a classifier from SVM and ResNET could yield the optimal performance for discriminating the geographical source area of coconut milk using FT-NIR. Furthermore, when using Micro-NIR, the classifier from LDA, SVM, KNN and ResNET delivered the highest accuracy. The performance discrimination models above were excellent when classified based on the kappa coefficient. This study concluded that both FT-NIR and Micro-NIR supported by classical to modern chemometric classifiers could be used to evaluate the geographical area source from coconut milk. Also, the method in this study includes a strategy for discovering feature-important NIR spectra for interpretability purposes, thereby facilitating the qualitative interpretation of results for all types of classifiers.

Recognition of Western Black-Crested Gibbon Call Signatures Based on SA_DenseNet-LSTM-Attention Network

As part of the ecosystem, the western black-crested gibbon (Nomascus concolor) is important for ecological sustainability. Calls are an important means of communication for gibbons, so accurately recognizing and categorizing gibbon calls is important for their population monitoring and conservation. Since a large amount of sound data will be generated in the process of acoustic monitoring, it will take a lot of time to recognize the gibbon calls manually, so this paper proposes a western black-crested gibbon call recognition network based on SA_DenseNet-LSTM-Attention. First, to address the lack of datasets, this paper explores 10 different data extension methods to process all the datasets, and then converts all the sound data into Mel spectrograms for model input. After the test, it is concluded that WaveGAN audio data augmentation method obtains the highest accuracy in improving the classification accuracy of all models in the paper. Then, the method of fusion of DenseNet-extracted features and LSTM-extracted temporal features using PCA principal component analysis is proposed to address the problem of the low accuracy of call recognition, and finally, the SA_DenseNet-LSTM-Attention western black-crested gibbon call recognition network proposed in this paper is used for recognition training. In order to verify the effectiveness of the feature fusion method proposed in this paper, we classified 13 different types of sounds and compared several different networks, and finally, the accuracy of the VGG16 model improved by 2.0%, the accuracy of the Xception model improved by 1.8%, the accuracy of the MobileNet model improved by 2.5%, and the accuracy of the DenseNet network model improved by 2.3%. Compared to other classical chirp recognition networks, our proposed network obtained the highest accuracy of 98.2%, and the convergence of our model is better than all the compared models. Our experiments have demonstrated that the deep learning-based call recognition method can provide better technical support for monitoring western black-crested gibbon populations.

Predicting Lead Changes in Tennis Matches based on Machine Learning Models

Based on men's singles data from the 2023 Wimbledon Tennis Championships, this study applies advanced statistical and machine learning techniques (including Principal Component Analysis PCA, Uniform Streaming Approximation and Projection UMAP, Random Forests, and Extreme Random Trees Extra-Trees) to explore the dynamics of lead changes during matches. By constructing mathematical models, the study predicts lead changes at key moments of the race that have a decisive impact on the race strategy and outcome. The experimental results show that the Extreme Random Trees model performs well on all evaluation metrics, especially in terms of accuracy, recall and F1 score, demonstrating its ability to handle complex interactions and variations in the data. In addition, the CatBoost model performed well in identifying actual leading changes, but with lower accuracy, possibly due to overfitting certain data patterns. The hybrid model, on the other hand, demonstrated a balanced performance, although it failed to outperform the extreme random tree model in terms of overall effectiveness.

Examining Quantiles in Structural Health Monitoring

A central task within structural health monitoring is to determine the condition of a given structure, such as bridges, by gathering and assessing data from integrated sensory systems. This data can be viewed as functional and subjected to corresponding statistical methodologies. Prompt detection of potential structural damage is crucial due to its potential impact on a structure's service life. One approach to address this challenge has been to use methods like (functional) principal component analysis ((F)PCA). FPCA identifies the primary directions of signal variability and monitors the evolution of corresponding scores over time. However, this conventional monitoring concept falls short in accounting for skewness or shifting effects, as it merely represents curves as deviations from the mean. A comprehensive analysis of the entire signals' distributions becomes imperative to tackle the multifaceted nature of structural damages and environmental influences adequately. In this innovative approach, we expand upon FPCA by incorporating a quantile perspective, where we model the significant variations in each sensor profile at various quantile levels. This empowers us to consider scores at different quantiles as vital monitoring metrics without imposing distributional assumptions on scores or error terms. Furthermore, our model takes into account confounding effects, particularly the temperature. We validate our method through simulation studies and real-data scenarios using strain and inclination data from long-term monitoring projects of German bridges. Our results demonstrate the effectiveness of our strategy as both a visual inspection tool and a reasonably accurate damage identification technique.

Hyperspectral Image Sharpening Using Fusion Techniques -A Case Study at Salah Al-Din Province/Iraq-

The data fusion process includes merging two or more pieces of information obtained from different sensors. Satellite image fusion research aims to create a new image by combining two images captured by different sensors using various methodologies. In this research, image sharpening tools were used to combine a hyperspectral image with a low spatial resolution captured by a Hyperion sensor mounted on the Earth Observation 1 (EO-1) satellite with a grayscale high spatial resolution image captured by Enhanced Thematic Mapper Plus (ETM +) sensor mounted on Landsat-8 (resampling first one to ensure equal spatial resolution of both images). In addition, three techniques were adopted for implementing the Fusion mechanism: the Principal Component Analysis PCA, the Nearest Neighbor Diffusion NNDifuse, and the Gram-Schmidt method; these were used to sharpen hyperspectral data using high spatial resolution. The result showed that the Gram-Schmidt method could give Hyperspectral images with higher spectral and spatial resolution in panchromatic image data more accurately than the other methods.

The Impact of Customer Intensity in Firm Performance: Tech Capabilities as A Moderating Variable

Entrepreneurial marketing “EM” is considered to be suitable for small businesses. However, since the perspective has shifted from product and sales to a market orientation, interest in the creation of customer value has grown and has been the subject of intense discussion. Thus, in this study, we aim to investigate and describe the significant relationship between Customer intensity “CI” and firm performance “FP” in SMEs operating in Khartoum-Sudan, by utilizing technological capabilities “TCPs” as a moderating variable. The resource-based view “RBV” provides the theoretical foundation for this study regarding the effect of CI on FP among TCPs. Regarding the study objective, we assumed the following hypothesis: H1: Customer intensity has a significant relationship with firm performance dimensions: {profitability, sustainability, and customer satisfaction}. H2: We assume that TCPs can positively moderate the relationship between CI and FP. Accordingly, this study is quantitative. Reliable with the purpose of this study. Furthermore, our study relied on the “Positivism philosophy”, a deductive approach to theory development. Therefore, overall, 255 responses were received from the community of SMEs in Khartoum. To analyze the data firstly the following principal component analysis PCA, Correlation, and Rotation matrix were utilized to test the appropriateness of the study structure “the pre-model” and check the validity of the questionnaire items. secondly, we used structural equation modeling (SEM) Path analysis to analyze and examine the significant relationships between all variables. The analysis revealed that customer intensity has a positive effect on profitability (p = 0.000), Customer intensity has also a positive effect on customer satisfaction (p = 0.001). Unlike, customer intensity has negative effects on sustainability (p = 0.216). Finally, the results show that technological capabilities positively moderate the relationship between Customer intensity and profitability (p = 0.000), sustainability (p = 0.007), and customer satisfaction (p = 0.000). Anyhow, these results confirmed the partial support for the first hypothesis and full support for the second hypothesis.

Seismic response prediction using a hybrid unsupervised and supervised machine learning in case of 3D RC frame buildings

The seismic vulnerability assessment represents an important step in monitoring the buildings’ capacity and checking their performance during and after earthquake events. The Nonlinear Time History Analysis (NL-THA) is considered the most reliable method that is used to calculate the exact structural behavior of any building. However, this sophisticated method is known for its complexity, the use of Finite Element (FE) software, and computational time consuming, especially in the case of tall buildings. For that reason, An Artificial Neural network (ANN) is used to develop a new model able to predict the essential Engineering Demand Parameters (EDPs), i.e., the Maximum Base Shear (MBS), the Maximum Inter-story Drift (MIDR) and the Maximum Roof Drift Ratio (RDR). Unsupervised algorithms such as the Principal Component Analysis PCA and the Autoencoder are coupled with the ANN to reduce the dimensionality, improve the dataset quality, and reduce the irrelevant features. More than 192,000 buildings are analyzed using the NL-THA and eighty artificial ground motions (GMs) to generate the dataset. The buildings’ characteristics are generated randomly from the selection range. The results showed that the Autoencoder-ANN model represents the highest performance compared to the PCA-ANN and ANN models. The Autoencoder-ANN model could quickly and accurately predict the seismic responses of unseen ground motions using only the building’s characteristics and the GM parameters without using any FE software.

Assessment of spatial distribution of lead in soils around an active military shooting range

Abstract In this study, the extent of Pb contamination around an active military shooting range is re-assessed to ascertain current contamination status comparatively to data’s reported 20 years ago. This is because lead bullet disintegration and mobility takes several years and extensively affects levels of soil-Pb contamination. A total of 120 topsoil samples were collected within the impact area (front) and non-impact area (back) for Pb determination. High concentrations of Pb were found at impact berm both front (28940±2996 μg/g) and back (775±128 μg/g). Spatial distribution of contamination reflects the distances from berm. Soil-Pb contamination around non-impact area was notable at back berm through to 100 m distance with significant difference in Pb, Cr, Ni and Zn levels. Concentration of Pb at berm was observed to have increase five-fold over a period of about twenty years from previous study with accumulation factor of about 1000. Principal component analysis PCA indicated 67 % of total metal load of range soil was majorly from impact areas of the berm. The correspondence analysis relay plot shows pollution order of Pb > Ni > Cd > Cr > Zn > Cu. This confirms soil contamination especially around the impact area, i.e. impact berm and firing lines and non-impact area at 0 m, 10 m, 50 m and 100 m. Environmental consequence of high soil-Pb levels within the range especially non-impact areas utilized for farming activities will leave much to desire. Hence, extensive and continuous monitoring is needed. However, remediation through appropriate soil washing technique could reduce Pb levels and improve soil condition regardless of age.

Optimization model for mineral composition data analysis and its application in jade classification

Abstract The classification of jade grade has always been a very critical part of the jade industry, and improving the accuracy of jade grade classification is of great significance to the sustainable development of the jade industry. The study constructs a mineral identification classification model based on Raman spectroscopy + PCA through Raman spectroscopy and PCA principal component analysis and analyzes the data of jade grades and constituents. The actual performance of this paper’s model is explored by comparing its effectiveness with other algorithmic models in jade classification and the accuracy of classification parameters. The model in this paper is feasible in classifying the four grades of Hetian jade (seed material, gobi material, shanliushui material, and shanmu material). Green dense jade’s main minerals are <unk>-quartz and a few other minerals, including albite, hematite, graphite, and tourmaline. The main compositions of the sample jade are SiO2, Al2O3, and K2O. The overall accuracy of this paper’s model in classifying Xinjiang Hotan jade grades is 97.9%, which is significantly higher than that of the KNN classification algorithm and SVM classification algorithm. The total accuracy of this paper’s model on each parameter of jade grade is 85, which is higher than the 60 of the KNN algorithm and the 62 of the SVM algorithm, and the classification accuracy grade is high.

WIFI-Based Human Identification of gait recognition in muti-scenario

With the development of 5G and the maturity of embedded technology, the Internet of Things has become the most promising technology at present, and human motion recognition and fingerprint feature recognition are hot research topics in the Internet of Things. At the same time, the way of human-computer interaction no longer satisfies the interaction only through screens and our computing devices. We hope to achieve human-computer interaction through simpler and more direct operations. For example, gestures, speech, etc., while the current methods of human recognition are mostly achieved through video or wearable devices, both of which have certain limitations. For example, when using video devices to recognize human bodies, it is necessary to ensure that there are no obstacles on the line of sight (LOS) in the video and sufficient ambient light is available. The Wi-Fi recognition method proposed in this article has the advantages of non-wearable, no light source restriction, and the ability to achieve human recognition using non-line-of-sight paths. Wi-Fi is ubiquitous in modern society, so using Wi-Fi signals for indoor human sensing has important research value. Based on existing Wi-Fi gait recognition work, this paper proposes a Wi-Fi-RSMID system for indoor human recognition. By analyzing the channel state information of Wi-Fi signals during human walking, the Wi-Fi-RSMID system extracts key feature information through PCA principal component analysis of thirty subcarriers of CSI. It collects 155 feature points from five principal components of PAC using a method that combines time-frequency domain information, and achieves human recognition through random subspace method. Experiments show that the system can effectively identify the identities of 15 people in different scenarios, with an average recognition rate of about 75.3% - 85.6%.

An Exploration of Face Recognition Methods Based on LBP Algorithm and PCA Analysis

Local Binary Pattern is an operator that can be effectively used for facial description, and its application field is relatively wide, which has made great contribution to the research and technical support related to face recognition. The traditional local texture feature extraction method of LBP operator is relatively single and cannot fully synthesize the problem of neighborhood correlation. Therefore, this paper proposes a face recognition method based on the combination of LBP algorithm and PCA principal component analysis method for the above problems. Through the application of this method, it can effectively weaken the influence of noise on the results and show the basic characteristic situation of the sub-block region, and the experiments of this method have confirmed that, compared with the use of PCA, LBP and other methods alone, the method in this study has a significant improvement in the recognition rate.

Contribution of socio-economic determinants to the management of sustainable performance: case study of Moroccan companies with Corporate Social Responsibility label

In a dynamic environment marked by significant shifts in production and marketing systems across various industries, companies are compelled to adhere to national and international standards to meet evolving demands from both internal and external stakeholders. These emerging requirements have spurred companies to adopt strategic initiatives, such as emphasizing Corporate Social Responsibility (CSR), as part of their managerial actions. The primary aim is to integrate two key management science themes: indicators derived from management control systems and the management of sustainable performance. Consequently, our focus lies in the socio-economic approach, particularly the socio-economic cost-value of activities, which influences the diversity of indicators on one hand and the management of sustainable performance on the other hand, mediated through their variety. This underscores the importance of managing sustainable performance by utilizing measures and data tied to socio-economic fundamentals, prompting us to formulate our research question as follows: "To what extent does the cost-value of activities impact the management of sustainable performance?" The empirical investigation transitions from simply describing variables within our sample companies to conducting factor analysis via Principal Component Analysis-PCA. This aims to establish a robust factorial structure and subject several constructs to testing, assessing the internal consistency of the model, the quality and reliability of the sample, and the validity of research hypotheses. Our findings align with prior research, indicating that diverse indicators can inform managers about social and environmental practices, fostering positive effects on sustainable firm performance. However, excessive reliance on these indicators may lead to confusion and dispersion within companies. Hence, it's imperative to tailor indicators to the specific context and objectives of each firm, considering industry-specific nuances. Our study demonstrates that leveraging a diverse range of indicators is pivotal for enhancing sustainable performance dimensions and managing socio-economic factors amidst external pressures. Evaluating the measurement model is crucial for validating the reliability and validity of measurement instruments, as well as assessing the structural model's overall quality, which includes tests like coefficient of determination-R², Goodness of fit-GoF, and others. It's important to note that our measurement model encompasses both reflective and formative constructs. The structural model's quality is gauged by its explanatory power, evaluated through R², indicating explained variance in the variables under scrutiny. To address this issue, we chose to examine the connections between key socio-economic factors and the management of sustainable performance in Moroccan companies certified with CSR labels. Through structural equation modeling, we found that effectively managing the cost-value of activities using diverse indicators plays a significant role in enhancing sustainable performance within CSR-certified companies.

A Profound Multitask System for Gender Identification face recognition, Confront Discovery, Point of interest Localization, and Head Position Estimation Hyperface

Machine learning is a technology that has risen in its usage and popularity in the last few years. A huge number of people from around the world are learning this technology and putting the knowledge to various use. Machine learning algorithms are capable of learning from the provided data with high accuracy. Even though a significant amount of research has been conducted on face recognition, the integrated model of face recognition, landmark localization, head posture estimation, and gender identification that is capable of high accuracy and speed has not yet been investigated. As a result, we have developed a face recognition system that can make predictions about photos that are comparable to those made by humans. The principal component analysis PCA and the SVM were used here to accomplish facial recognition. In feature extraction, to reduce the dimensionality of large datasets, principal component analysis is performed. After the data have been preprocessed, they are entered into the SVM classifier to be used for image classification. The study of this is done via visualization, and it is used to measure the effectiveness of the model. This face recognition algorithm has an accuracy of at least 80% when it comes to classifying people's portraits. The findings of the experiments show that the suggested technique can successfully identify faces since it employs a feature-based algorithm that combines PCA classification and SVM detection.

SAT065 Developmental Programming: Sex-specific Programming Of Pancreas In A Sheep Model Of Polycystic Ovary Syndrome

Abstract Disclosure: K.M. Halloran: None. A.K. Vyas: None. N. Saadat: None. V. Padmanabhan: None. Prenatal testosterone (T) treatment from gestational days (GD) 30-90 (sexually dimorphic window of fetal development where male fetuses naturally see elevated T) leads to reproductive and metabolic disruptions in female sheep that includes dyslipidemia, insulin resistance, and compensatory hyperinsulinemia similar to that seen in women with PCOS. The effects of excess T are seen as decreased pancreatic/fetal weight in female fetuses as opposed to an increase in beta cell size and lipid/collagen accumulation in adult females. We addressed the prenatal T-induced gene expression changes contributing to pancreatic compromise to determine if 1) the programmed event persists after cessation of T treatment, 2) T females (TF) are similar to control males (CM), and 3) it is exacerbated in the T males (TM). Time-mated pregnant sheep were administered T propionate (100 mg in 2 ml corn oil) or vehicle between GD 30-90 to obtain control female (CF), TF, CM, and TM fetuses (n=6/group). On ∼GD 120 (term: 147), fetuses were weighed and pancreas was collected, weighed, and the tail portion stored at -80°C until next-generation sequencing for gene expression profiling to identify potential pathways influenced by T and fetal sex. After removal of outliers (determined by Principal Component Analysis-PCA using SIMCA), 3D PCA plots showed clear separation of the treatment groups after stratifying by fetal sex. DESeq2 R package was used to identify differentially expressed genes (DEGs) between the comparisons of interest (CF vs TF, TF vs CM, and CM vs TM). Overall, T fetuses weighed less (P<0.05). Cohen’s effect size analysis found a large magnitude (Cohen’s d>0.8) decrease in fetal weight for TF vs CF, TM vs CM, and TF vs CM and a medium magnitude (d=0.5) decrease in pancreas weight for TF vs CM. Pancreatic:fetal weight ratios did not differ by treatment or fetal sex. 16 DEGS (1 up, 15 down) were identified between TF and CF, 7 DEGs (2 up, 4 down) between TF and CM, and 2 DEGs (up) between TM and CM (Adj P<0.1) with no overlap amongst them, suggesting that the effect of T in females is distinct from males and that TF are more similar to CM. Preliminary analysis indicates that the majority of the DEGs affected in TF are related to immune-signaling pathways, including c-type lectin domain family 4 member G (CLEC4G), cluster of differentiation 52 (CD52), and interleukin 21 receptor (IL21R); CD52, IL21R, and members of the CLEC4 family are implicated in autoimmune pathways linked to Type 1 diabetes. The 2 DEGs affected in TM are involved in plasma membrane dynamics: fer-1 like family member 6 (FER1L6) and chloride voltage gated channel 2 (CLCN2). These changes evident at the gene expression level a month after the cessation of T treatment is suggestive of sexually dimorphic reprogramming of fetal pancreas by exposure to excess T. Further, the endogenous increases in T seen in males may at least in part be responsible for sexually dimorphic programming. Presentation: Saturday, June 17, 2023

Model for Developing the Creativity of Future Engineers

The article presents factors and their interrelations that influence the development of engineers’ creativity. The subject of the article is in the field of human resources management and was in-spired by the results of research related to the analysis of the competency potential of future employees. The main goals of this paper are to consider how can competencies of future engineers in the area of creativity be developed and what does creativity depend on – internal and external factors; can a model of enhancing creativity of future engineers be worked out? The research was carried out between 2014-2020, on a group of 861 students. The survey method was used, and the research tool was a self-evaluation questionnaire. The research used regression analyses, Spearman’s rank correlation coefficients, the power of interrelation between chosen variables (social competencies and forms of education) was evaluated. Hierarchical multiple regression was carried out to verify the objective of the research. The analysis of correlations was verified and visualised with principal components analysis PCA, as well with cluster analysis using Ward’s method. The main value of the article is the developed graphic model that reflects the stages of creative development as a competence that is a very important competence resource of the company. A mathematical model that includes the analysis, based on hierarchical multiple regression, revealed that social competencies and forms of learning examined are strongly correlated with 'creative thinking' competence, and this correlation is statistically significant.

Effective full connection neural network updating using a quantized full FORCE algorithm

This paper presents a new training algorithm that can update the situation of layers’ network, and therefore, connections, neurons, and firing rate of neurons based on FORCE (first-order reduced and controlled error) training algorithm. The Quantized Full FORCE algorithm (QFF) also updates the number of neurons and connections between different layers in the network per iteration in a way that the whole firing rate of each layer is updated via selecting the best neurons and combining strong features. The update method is sequential, so that with each instance passing through the network, the network structure is updated with the Full FORCE algorithm. The algorithm updates the structure of networks with a multiple/single middle layer of the supervised version of feed forward networks such as Multilayer perceptron (MLP), changing them into partially-connected networks. A combination of principal component analysis PCA and Linear Discriminant Analysis (LDA) algorithms has been used to cluster the network input features. The paper focuses on the deep supervised MLP network with backpropagation (BP) and various datasets and its comparison with other MLP based stat of art methods and hybrid evolutionary algorithms. We achieved 98.15 percent accuracy for facial expression 98.6 and 97.7 percent for Wisconsin breast Cancer and Iris Flower in respectively. The training algorithm employed in the study enjoys a lower computational complexity while yielding faster and more accurate convergence, starting with a very low level of errors of 0.009 in comparison with the full connection network and it solves the challenge of getting stuck in local minima and poor convergence of Gradient Decent with BP.

Optimized variable selection and machine learning models for olive oil quality assessment using portable near infrared spectroscopy

Olive oil is a key component of the Mediterranean diet, rich in antioxidants and beneficial monounsaturated fatty acids. As a result, high-quality olive oil is in great demand, with its price varying depending on its quality. Traditional chemical tests for assessing olive oil quality are expensive and time-consuming. To address these limitations, this study explores the use of near infrared spectroscopy (NIRS) in predicting key quality parameters of olive oil, including acidity, K232, and K270. To this end, a set of 200 olive oil samples was collected from various agricultural regions of Morocco, covering all three quality categories (extra virgin, virgin, and ordinary virgin). The findings of this study have implications for reducing analysis time and costs associated with olive oil quality assessment. To predict olive oil quality parameters, chemical analysis was conducted in accordance with international standards, while the spectra were obtained using a portable NIR spectrometer. Partial least squares regression (PLSR) was employed along with various variable selection algorithms to establish the relationship between wavelengths and chemical data in order to accurately predict the quality parameters. Through this approach, the study aimed to enhance the efficiency and accuracy of olive oil quality assessment. The obtained results show that NIRS combined with machine learning accurately predicted the acidity using iPLS methods for variable selection, it generates a PLSR with coefficients of determination R2 = 0.94, root mean square error RMSE = 0.32 and ratios of standard error of performance to standard deviation RPD = 4.2 for the validation set. Also, the use of variable selection methods improves the quality of the prediction. For K232 and K270 the NIRS shows moderate prediction performance, it gave an R2 between 0.60 and 0.75. Generally, the results showed that it was possible to predict acidity K232, and K270 parameters with excellent to moderate accuracy for the two last parameters. Moreover, it was also possible to distinguish between different quality groups of olive oil using the principal component analysis PCA, and the use of variable selection helps to use the useful wavelength for the prediction olive oil using a portable NIR spectrometer.

Comparison of the Metabolic Profile between Unstimulated and Stimulated Saliva Samples from Pregnant Women with/without Obesity and Periodontitis.

This study aimed to compare the metabolic profile of unstimulated (US) and stimulated (SS) saliva samples from pregnant women with/without obesity and periodontitis. Ninety-six pregnant women were divided into: obesity + periodontitis (OP = 20); obesity/no periodontitis (OWP = 27); normal BMI + periodontitis (NP = 20); and normal BMI/no periodontitis (NWP = 29). US and SS samples were collected by expectoration and chewing of sterilized parafilm gum, respectively, and samples were individually analyzed by Proton Nuclear Magnetic Resonance (1H-NMR). Univariate (t test and correlations) and multivariate (Principal Component Analysis-PCA, and Partial Least Square-Discriminant Analysis-PLS-DA with Variance Importance Projection-VIP scores) and Metabolite Set Enrichment Analysis were done (p < 0.05). Metabolites commonly found in all groups in elevated concentration in US samples were 5-Aminopentoate, Acetic acid, Butyric acid, Propionic acid, Pyruvic acid, and Succinic acid. They were mainly related to the butyrate metabolism, citric acid cycle, amino sugar metabolism, fatty acids biosynthesis, pyruvate metabolism, glutamate metabolism, and Warburg effect. Metabolites commonly found in all groups that were in elevated concentration in SS samples were Citrulline, Fumaric acid, Histidine, N-acetyl glutamine, N-acetylneuraminic acid, para-hydroxyphenylacetic acid, Proline, Tyrosine. Although some differences were found between unstimulated and stimulated saliva samples from pregnant women with/without obesity and periodontitis, stimulated saliva collection seems adequate, demonstrating similar metabolic pathways to unstimulated saliva samples when groups are compared.

The Evaluation of the Grade of Leaf Disease in Apple Trees Based on PCA-Logistic Regression Analysis

Extensive research suggested that the core of how to use pesticides scientifically is the careful and accurate determination of the severity of crop diseases. The existing grading standards of plant leaf diseases have been excessively singular. Thus, the diseases roughly fall into general and severe grades. To address the above problems, this study considered the effect of the distribution of disease spots, and two evaluation indicators (termed the imbalance degree and main vein distance) were newly added to optimize the grading criteria of apple leaf diseases. Combined with other factors, the grade evaluation indicator was determined through PCA principal component analysis. A gradual multivariate logistic regression algorithm was proposed to evaluate apple leaf disease grade and an optimized apple leaf disease grade evaluation model was built through PCA-logistic regression analysis. In addition, three common apple leaf diseases with a total of 4500 pictures (i.e., black rot, scab, and rust) were selected from several open-source datasets as the subjects of this paper. The object detection algorithm was then used to verify the effectiveness of the new model. As indicated by the results, it can be seen from the loss curve that the loss rate reaches a stable range of around 70 at the epoch. Compared with Faster R-CNN, the average accuracy of Mask R-CNN for the type and grade recognition of apple leaf disease was optimized by 4.91%, and the average recall rate was increased by 5.19%. The average accuracy of the optimized apple leaf disease grade evaluation model was 90.12%, marking an overall increase of 20.48%. Thus, the effectiveness of the new model was confirmed.