Holdout Technique Research Articles

Machine learning-based modelling of zenith wet delay using terrestrial meteorological data in the Brazilian territory

Abstract. The Zenith Total Delay (ZTD) is one of the primary error sources derived from the neutral atmosphere associated with the GNSS (Global Navigation Satellite Systems) technique. Zenith Wet Delay (ZWD) is the smallest part of the ZTD, but the high variability is caused by spatial-temporal variation, making the modelling of this component a challenging task. Although ZWD is considered an error in GNSS positioning, it is also a variable composed mainly of water vapour and can, therefore, be used for atmospheric investigations, and assists in climate studies for precipitation events. In this work, a model was trained to estimate the delay wet component from surface atmospheric parameters. The training data comes from 29 radiosonde stations around Brazil, for a six-year period (2017 to 2022), with data collected at 12 h UTC (Universal Time Coordinated). The model was validated using the holdout technique, with 70% of the data used in training and 30% for validation (cross-validation analysis). The generated model achieved a RMSE (Root Mean Squared Error) of approximately 38 mm, with an 81% of determination coefficient.

Evaluation of the Effectiveness of Neural Network Models for Analyzing Customer Review Sentiments on Marketplace

According to the 2019 report, Tokopedia is the most visited marketplace with 140,000,000 visitors per month, making it one of the most popular marketplaces in Indonesia. Customers have the opportunity to write reviews about the products they purchase at the end of the transaction process on Tokopedia. The aim of this research is to conduct sentiment analysis on product reviews on Tokopedia. Three neural networks that will be used for text classification are Bi-GRU, GRU, and LSTM. The data processing technique is divided into training and testing samples, split into 80%:20% using the holdout technique. The BI-GRU algorithm has an accuracy of 0.93% and precision of 0.96, better than the other two methods LSTM and GRU, which each have an accuracy of 0.92 and recall of 0.91.

Survival analysis of recurrent breast cancer patients using mix Bayesian network

IntroductionBreast cancer (BC) is the most common cancer among women. Iranians have an 11% BC recurrence rate, which lowers their survival rates. Few studies have investigated cancer recurrence survival rates. This study's major purpose is to use a mixed Bayesian network (BN) to analyze recurrent patients' survival. Material and methodsThis study aimed to evaluate the pathobiological features, age, gender, final status, and survival time of the patients. Bayesian imputation was used for missing data. The performance of BN was optimized through the utilization of a blacklist and prior probability. After structural and parametric learning, posterior conditional probabilities and mean survival periods for the node arcs were predicted. The hold-out technique based on the posterior classification error was used to investigate the model's validation. ResultsThe study included 220 cancer recurrence patients. These patients averaged 47 years old. The BN with a blacklist and prior probability has a higher network score than other networks. The hold-out technique verified structural learning. The Directed Acyclic Graph showed a statistically significant relationship between cancer biomarkers (ER, PR, and HER2 receptors), cancer stage, and tumor grade and patient survival duration. Patient death was also significantly associated with education, ER, PR, HER2, and tumor grade. The BN reports that HER2 negative, ER positive, and PR positive patients had a higher survival rate. ConclusionSurvival and death of relapsed patients depend on biomarkers. Based on the findings, patient survival can be predicted with their features.

Enhanced the prediction approach of diabetes using an autoencoder with regularization and deep neural network

Diabetes mellitus is considered one of the foremost common and extreme diseases worldwide. A precise and early diagnosis of diabetes is essential to avoid complications and is of crucial importance to the medical care that patients get. To achieve that, we need to develop a model to predict diabetes. There are many prediction models, but they suffer from some problems such as the accuracy of prediction being poor and the time complexity. The prediction process is highly dependent on important features. So, in this paper, we proposed a new model called (CAER-DNN) that depends on an unsupervised technique for generating newly important features and a deep neural network for the prediction process. The unsupervised technique is called complete autoencoder with regularization techniques (CAER) that uses to reconstruct the original features (newly learned features). It is focused too much on training the most important learned features and misses out on less important features. Thus, improving the performance of the prediction process. These important features are used as input to the deep neural network for the prediction of diabetes. Our model is applied to two sets of data including Pima Indian and Mendeley diabetic datasets. Based on the 10-fold cross-validation technique Pima Indian dataset achieves high performance in evaluation measures (f1-score 97.38%, accuracy, recall 97.25%, specificity 97.59%, precision 97.53%,). While the Mendeley diabetes dataset achieved high performance in evaluation measures (f1-score 94.51%, accuracy 98.48, recall 91.74%, accuracy-balance 98.21%, precision 98.21%) based on the holdout technique. compared with other existing machine learning and deep learning techniques our model outperformed existing techniques.

A novel dynamic Bayesian network approach for data mining and survival data analysis

BackgroundCensorship is the primary challenge in survival modeling, especially in human health studies. The classical methods have been limited by applications like Kaplan–Meier or restricted assumptions like the Cox regression model. On the other hand, Machine learning algorithms commonly rely on the high dimensionality of data and ignore the censorship attribute. In addition, these algorithms are more sophisticated to understand and utilize. We propose a novel approach based on the Bayesian network to address these issues.MethodsWe proposed a two-slice temporal Bayesian network model for the survival data, introducing the survival and censorship status in each observed time as the dynamic states. A score-based algorithm learned the structure of the directed acyclic graph. The likelihood approach conducted parameter learning. We conducted a simulation study to assess the performance of our model in comparison with the Kaplan–Meier and Cox proportional hazard regression. We defined various scenarios according to the sample size, censoring rate, and shapes of survival and censoring distributions across time. Finally, we fit the model on a real-world dataset that includes 760 post gastrectomy surgery due to gastric cancer. The validation of the model was explored using the hold-out technique based on the posterior classification error. Our survival model performance results were compared using the Kaplan–Meier and Cox proportional hazard models.ResultsThe simulation study shows the superiority of DBN in bias reduction for many scenarios compared with Cox regression and Kaplan–Meier, especially in the late survival times. In the real-world data, the structure of the dynamic Bayesian network model satisfied the finding from Kaplan–Meier and Cox regression classical approaches. The posterior classification error found from the validation technique did not exceed 0.04, representing that our network predicted the state variables with more than 96% accuracy.ConclusionsOur proposed dynamic Bayesian network model could be used as a data mining technique in the context of survival data analysis. The advantages of this approach are feature selection ability, straightforward interpretation, handling of high-dimensional data, and few assumptions.

A Review on Data Mining Technology

Presently days the method involved with removing significant principles from large and complex data are called data mining. Data mining has a rising fame in each field today. Data units are laid out in client situated enterprises, for example, showcasing, money and telecom to deal with the client beat and obtaining, specifically. Among the data mining techniques, order calculations are utilised in examinations led for client securing to anticipate the likely clients of the organization being referred to in the connected business. The idea of Data Mining has arisen to meet the prerequisite of speedy and exact data support for dynamic cycle. Data Mining is investigation of data to distinguish connection between various data components or substances. The course of data mining can likewise include connection or relationship between at least two data components, substances or occasions. They permit associations to make proactive, information-driven choices and answer questions that were already excessively tedious to determine. In this exploration we have zeroed in on the correlation of different Data Mining apparatuses which are useful and set apart as the significant field of data mining Technologies. As we know that numerous public and Multinational organisations and little or enormous associations are worked in better places of the various nations. Every activity might produce large data or unstructured. This sort of huge data is accessible as byte which has definitely changed in different regions. To examine, oversee and pursue a choice of such sort of colossal measure of data we really want strategies called the data mining which will changing in many fields. Bank advertising data set in UCI Machine Learning Data Set was utilised by making models with similar grouping calculations in various data mining programs. Exactness, accuracy and measure measures were utilised to test exhibitions of the characterisation models. While making the grouping models, the test and preparing data sets were arbitrarily separated by the holdout technique to assess the exhibition of the data set.

SQL Injection Attack Detection Using Machine Learning Algorithm

The structured query language injection attack (SQLIA) is a well-known cyberattack targeting vulnerabilities in web-based applications; it is used to carry out illegal information control language, bypass confirmation measures, and get access to restricted data. There was some consideration given to existing systematic reviews in the literature. Contemporary systematic reviews frequently incorporate both older and more contemporary works in the topic. Therefore, we restricted ourselves to recently published works. For the current study, I used information from 2012 through 2020. Encryption, XML, design coordination, parsing, and machine learning are just some of the methods and systems that can be used to spot and prevent SQL injection attacks. The Machine Learning (ML) process, which has been proved to be important for SQLIA relief, is applied with the help of guarded coding. Machine learning approaches require a large amount of data for model preparation and only handle a few number of attack types. The use of ML methods may alleviate a particularly challenging vision impairment SQL injection attack. In the Waikato Climate for Data Exploration study, we looked at the following methods: Logistic Regression (LRN), Stochastic Gradient Descent (SDG), Sequential Minimal Optimization (SMO), Bayes Network (BNK), Instance-Based Learner (IBK), Multilayer Perceptron (MLP), Naive Bayes (NB), and J48. Wait (70%) and 10-fold Cross Validation assessment procedures were used to survey the presentation of the regulated learning grouping calculations to choose the optimal calculation. Accuracy values for SMO, IBK, and J48 were found to be 98.7785%, 98.4285%, and 98.2985% using the Cross Validation method, and 98.7956%, 98.1526%, and 100 using the Hold-Out method. Using the Cross Validation method SMO took IBK and J48 10.15 seconds, 0.06 seconds, and 14.12 seconds, whereas using the Hold-Out method SMO took 9.71 seconds, 0.16 seconds, and 14.28 seconds to construct their models. Based on the findings, IBK was chosen as the classifier for SQLIA detection and prevention since it was the fastest to train a model using the Cross Validation strategy and had the best overall performance. Not only is accuracy essential when choosing an algorithm for predictive analytics, but also a variety of performance assessment indicators.

IoT Device for Sitting Posture Classification Using Artificial Neural Networks

Nowadays, the percentage of time that the population spends sitting has increased substantially due to the use of computers as the main tool for work or leisure and the increase in jobs with a high office workload. As a consequence, it is common to suffer musculoskeletal pain, mainly in the back, which can lead to both temporary and chronic damage. This pain is related to holding a posture during a prolonged period of sitting, usually in front of a computer. This work presents a IoT posture monitoring system while sitting. The system consists of a device equipped with Force Sensitive Resistors (FSR) that, placed on a chair seat, detects the points where the user exerts pressure when sitting. The system is complemented with a Machine Learning model based on Artificial Neural Networks, which was trained to recognize the neutral correct posture as well as the six most frequent postures that involve risk of damage to the locomotor system. In this study, data was collected from 12 participants for each of the seven positions considered, using the developed sensing device. Several neural network models were trained and evaluated in order to improve the classification effectiveness. Hold-Out technique was used to guide the training and evaluation process. The results achieved a mean accuracy of 81% by means of a model consisting of two hidden layers of 128 neurons each. These results demonstrate that is feasible to distinguish different sitting postures using few sensors allocated in the surface of a seat, which implies lower costs and less complexity of the system.

Opinion Analysis for Emotional Classification on Emoji Tweets using the Naïve Bayes Algorithm

Opinion Analysis is a research study needed to social media, since the content could become a trending topic and has a significant impact on social life. One of the social media that have a big contribution to cyberspace and information development is Twitter. In the Twitter application, users can insert images that represent emotions, facial expressions, or icons. Emoji is a graphic symbol in the form of an image to express a thing, with the Emoji, a text can be read and understood according to its meaning because the image represents it. Of the several things that have been mentioned then, the researchers conducted research on the classification of tweet content based on the use of Emojis. This study aims to determine the emotional uses of Twitter in one period. Every tweet on the Twitter timeline, which contains both text and Emojis, will be classified according to several categories. The algorithm used was Naive Bayes. It calculated the probability of Emoji tweet to obtain the text classification with Emojis. The results of the classification of emotions are grouped with three categories, namely angry, joy, and sad, it showed that the category joy had become the emotional trend of Twitter users where Emojis (x1f60a) dominate the most. Meanwhile, the accuracy of the algorithm used to reach 90% with a 70:30 holdout technique.

An Efficient and Effective Approach for Flooding Attack Detection in Optical Burst Switching Networks

Optical burst switching (OBS) networks are frequently compromised by attackers who can flood the networks with burst header packets (BHPs), causing a denial of service (DoS) attack, also known as a BHP flooding attack. Nowadays, a set of machine learning (ML) methods have been embedded into OBS core switches to detect these BHP flooding attacks. However, due to the redundant features of BHP data and the limited capability of OBS core switches, the existing technology still requires major improvements to work effectively and efficiently. In this paper, an efficient and effective ML-based security approach is proposed for detecting BHP flooding attacks. The proposed approach consists of a feature selection phase and a classification phase. The feature selection phase uses the information gain (IG) method to select the most important features, enhancing the efficiency of detection. For the classification phase, a decision tree (DT) classifier is used to build the model based on the selected features of BHPs, reducing the overfitting problem and improving the accuracy of detection. A set of experiments are conducted on a public dataset of OBS networks using 10-fold cross-validation and holdout techniques. Experimental results show that the proposed approach achieved the highest possible classification accuracy of 100% by using only three features.

Prediction of Chlorophyll-a Concentrations in the Nakdong River Using Machine Learning Methods

Many studies have attempted to predict chlorophyll-a concentrations using multiple regression models and validating them with a hold-out technique. In this study commonly used machine learning models, such as Support Vector Regression, Bagging, Random Forest, Extreme Gradient Boosting (XGBoost), Recurrent Neural Network (RNN), and Long–Short-Term Memory (LSTM), are used to build a new model to predict chlorophyll-a concentrations in the Nakdong River, Korea. We employed 1–step ahead recursive prediction to reflect the characteristics of the time series data. In order to increase the prediction accuracy, the model construction was based on forward variable selection. The fitted models were validated by means of cumulative learning and rolling window learning, as opposed to the hold–out technique. The best results were obtained when the chlorophyll-a concentration was predicted by combining the RNN model with the rolling window learning method. The results suggest that the selection of explanatory variables and 1–step ahead recursive prediction in the machine learning model are important processes for improving its prediction performance.

Big Data Sentimental Analysis Using Document to Vector and Optimized Support Vector Machine

With the rapid evolution of the internet, using social media networks such as Twitter, Facebook, and Tumblr, is becoming so common that they have made a great impact on every aspect of human life. Twitter is one of the most popular micro-blogging social media that allow people to share their emotions in short text about variety of topics such as company’s products, people, politics, and services. Analyzing sentiment could be possible as emotions and reviews on different topics are shared every second, which makes social media to become a useful source of information in different fields such as business, politics, applications, and services. Twitter Application Programming Interface (Twitter-API), which is an interface between developers and Twitter, allows them to search for tweets based on the desired keyword using some secret keys and tokens. In this work, Twitter-API used to download the most recent tweets about four keywords, namely, (Trump, Bitcoin, IoT, and Toyota) with a different number of tweets. “Vader” that is a lexicon rule-based method used to categorize downloaded tweets into “Positive” and “Negative” based on their polarity, then the tweets were protected in Mongo database for the next processes. After pre-processing, the hold-out technique was used to split each dataset to 80% as “training-set” and rest 20% “testing-set.” After that, a deep learning-based Document to Vector model was used for feature extraction. To perform the classification task, Radial Bias Function kernel-based support vector machine (SVM) has been used. The accuracy of (RBF-SVM) mainly depends on the value of hyperplane “Soft Margin” penalty “C” and γ “gamma” parameters. The main goal of this work is to select best values for those parameters in order to improve the accuracy of RBF-SVM classifier. The objective of this study is to show the impacts of using four meta-heuristic optimizer algorithms, namely, particle swarm optimizer (PSO), modified PSO (MPSO), grey wolf optimizer (GWO), and hybrid of PSO-GWO in improving SVM classification accuracy by selecting the best values for those parameters. To the best of our knowledge, hybrid PSO-GWO has never been used in SVM optimization. The results show that these optimizers have a significant impact on increasing SVM accuracy. The best accuracy of the model with traditional SVM was 87.885%. After optimization, the highest accuracy obtained with GWO is 91.053% while PSO, hybrid PSO-GWO, and MPSO best accuracies are 90.736%, 90.657%, and 90.557%, respectively.

A novel machine learning based feature selection for motor imagery EEG signal classification in Internet of medical things environment

Abstract In Internet of Medical Things (IoMT) environment, feature selection is an efficient way of identifying the most discriminant health-related features from the original feature-set. Feature selection not only finds the best informative features, but also helps in reducing the overall dimensions of the given dataset. In this paper, the actual feature-set is obtained from Brain Computer Interface (BCI) Competition-II Dataset-III motor-imagery electroencephalogram (EEG) signal using the Adaptive Auto-regressive (AAR) feature extraction technique. Based on the order (number of AR coefficients) of the AAR algorithm, two variants of datasets have been generated: 12 ( o r d e r = 6 per electrode) and 24 ( o r d e r = 12 per electrode) AAR features datasets. Here, a new fuzzified version of discernibility matrix has been proposed to determine a subset of features, which provides the best classification accuracy. In order to find the best feature subset, various types of dissimilarity measures have been used and compared with one another in our proposed fuzzy discernibility matrix (FDM) based feature selection technique. We have implemented the proposed algorithm on the given datasets using both the holdout technique as well as the 10-fold cross-validation in our study. The performances of the selected feature-subsets are evaluated based on accuracies using the Support Vector Machine (SVM) and Ensemble variants of classifiers. The empirical results obtained from our experiments in this paper is competitive in terms of accuracy and outperformed the other popular t-test, Kullback–Leibler Divergence (KLD), Bhattacharyya distance and Gini index based feature selection techniques. Our proposed FDM based feature selection algorithm using holdout technique provides 80% and 78.57% accuracies for the 12 and 24 features AAR datasets respectively. The results obtained in the holdout technique with only 50% of the best discriminant features are even better than the performances obtained while using the original feature-sets (without using any feature selection technique). Again, it gives 78.57% and 75.57% mean-accuracies from 5 × 10-fold cross-validations using only 6and 12 most discriminant AAR features from the actual 12 & 24 features-sets respectively.

A Hybrid Feature Extraction Method With Regularized Extreme Learning Machine for Brain Tumor Classification

Brain cancer classification is an important step that depends on the physician’s knowledge and experience. An automated tumor classification system is very essential to support radiologists and physicians to identify brain tumors. However, the accuracy of current systems needs to be improved for suitable treatments. In this paper, we propose a hybrid feature extraction method with a regularized extreme learning machine (RELM) for developing an accurate brain tumor classification approach. The approach starts by preprocessing the brain images by using a min–max normalization rule to enhance the contrast of brain edges and regions. Then, the brain tumor features are extracted based on a hybrid method of feature extraction. Finally, a RELM is used for classifying the type of brain tumor. To evaluate and compare the proposed approach, a set of experiments is conducted on a new public dataset of brain images. The experimental results proved that the approach is more effective compared with the existing state-of-the-art approaches, and the performance in terms of classification accuracy improved from 91.51% to 94.233% for the experiment of the random holdout technique.

The analysis of the effects of acute rheumatic fever in childhood on cardiac disease with data mining

BackgroundAcute rheumatic fever (ARF) is an important disease that is frequently seen in Turkey, it is necessary to develop solutions to cure the disease. It is believed that new data analysis methods may be applied to this disease, and this may be useful to discover previously unrecognized patterns. Data mining of existing records and data repositories may improve knowledge on the diagnosis and management of ARF. In this regard, we planned to make a contribution to the development of new solutions by approaching the problem from a different standpoint. ObjectivesThe aim of this study is to analyse the effects of ARF undergone during childhood on the basis of cardiac diseases by using data mining methods. Materials and methodsClassification methods of data mining were used, and experiments were conducted on five algorithms. The records of the patients diagnosed with ARF were analysed by setting models with naive Bayes classifier, decision trees (CART, C4.5, C5.0, C5.0 boosted) and random forest algorithms. The performances of the algorithms that were derived were then compared. Among model performance evaluation techniques, the hold-out, cross-validation and bootstrap methods were tested in diverse ways in an applied manner. Within the scope of the research, the dataset comprising records of 297 patients was utilised in cooperation with İstanbul Medeniyet University Göztepe Training and Research Hospital’s Pediatric Cardiology Clinic (İstanbul Medeniyet Üniversitesi Göztepe Eğitim ve Araştırma Hastanesi Çocuk Kardiyolojisi Kliniği). Data analysis was carried out with the data of the remaining 201 patients following pre-processing. ResultsThe results that were obtained from different algorithms were compared based on the model performance evaluation criteria. The best result was shown under the CART model by using the hold-out technique (80% training, 20% testing). According to this model, the importance values of the predictive attributes were listed, and it was found that the “teleNormal” and “cardiomegaly” attributes were not required for ARF diagnosis and treatment. In compliance with this result, it was thought that it should not be necessary for patients have a chest x-ray which is needed for diagnosis of “teleNormal” and “cardiomegaly”. This will help reduce costs and thus contribute to the health economy while preventing patients from having unnecessary x-rays. Discussion and conclusionThe results of this study showed that data mining techniques may be used to analyse diseases such as ARF. The important attributes that affect the disease were obtained in accordance with the results. The results of the best model (CART) may be broadened in numerous ways and provide information for both experienced and inexperienced physicians. This study is considered to be significant as it helps data mining methods become more prevalently used for data analysis in fields of medicine and healthcare.

The importance of climatic factors and outliers in predicting regional monthly campylobacteriosis risk in Georgia, USA

Incidence of Campylobacter infection exhibits a strong seasonal component and regional variations in temperate climate zones. Forecasting the risk of infection regionally may provide clues to identify sources of transmission affected by temperature and precipitation. The objectives of this study were to (1) assess temporal patterns and differences in campylobacteriosis risk among nine climatic divisions of Georgia, USA, (2) compare univariate forecasting models that analyze campylobacteriosis risk over time with those that incorporate temperature and/or precipitation, and (3) investigate alternatives to supposedly random walk series and non-random occurrences that could be outliers. Temporal patterns of campylobacteriosis risk in Georgia were visually and statistically assessed. Univariate and multivariable forecasting models were used to predict the risk of campylobacteriosis and the coefficient of determination (R 2) was used for evaluating training (1999–2007) and holdout (2008) samples. Statistical control charting and rolling holdout periods were investigated to better understand the effect of outliers and improve forecasts. State and division level campylobacteriosis risk exhibited seasonal patterns with peaks occurring between June and August, and there were significant associations between campylobacteriosis risk, precipitation, and temperature. State and combined division forecasts were better than divisions alone, and models that included climate variables were comparable to univariate models. While rolling holdout techniques did not improve predictive ability, control charting identified high-risk time periods that require further investigation. These findings are important in (1) determining how climatic factors affect environmental sources and reservoirs of Campylobacter spp. and (2) identifying regional spikes in the risk of human Campylobacter infection and their underlying causes.

Quantifying physical activity via pedometry in elementary physical education.

The objective of this study was to determine a pedometer steps per minute standard for quantifying the lesson time that first- and second-grade physical education students spent in moderate to vigorous physical activity (MVPA). The sample was divided into validation (N = 246) and cross-validation (N = 123) samples using the holdout technique. Using the criterion test model, steps per minute cut points were empirically and judgmentally determined. C-SOFIT systematic observation was the criterion instrument and pedometry was the predictor instrument. Data were collected from 45 physical education lessons implemented in six schools. The three-step analytic procedure of computing mastery/nonmastery outcome probabilities, phi coefficients, and error proportions was used to determine the optimal steps per minute cut point for quantifying 33.33% of the physical education lesson time engaged in MVPA within a 30 class. Steps per minute was highly correlated with observation (r = 0.74-0.86, P < 0.0001). Five steps per minute scores that were accurate indicators of 33.33% of the class time engaged in MVPA in the validation sample were accurate indicators of steps per minute in the cross-validation sample. The optimal steps per minute cut point ranged from 60.00 to 63.00, which is equivalent to 1800-1890 steps in a 30-min physical education class. Data supports the use of pedometry steps per minute values as an accurate indicator of MVPA. Pedometry demonstrates promise as a viable large-scale surveillance instrument for measuring MVPA in physical education.

Aperture antenna shape prediction by feedforward neural networks

The emergence of adaptive "smart" materials has led to the design of active aperture antennas. Inherent in these antennas is the ability to change their shape in real time to meet various performance characteristics. When examining the usefulness of these antennas, one of the primary concerns is the antenna shape needed for a particular radiation pattern. Aperture antenna shape prediction is also a concern in the industrial production of semi-paraboloidal antennas. The work in this study employs an artificial neural network to model the aperture antenna shape in real time. To test the accuracy of the network, the "threefold holdout technique" was employed. In this technique, sets of examples are "held out" of the training process and used to obtain the "true error" of the network. The network accurately predicted the aperture shape exactly, to within three significant digits, 96% of the time.