Decision Tree Classification Algorithm Research Articles

Classification and Regression Tree Models for Remote Recognition of Black and Odorous Water Bodies Based on Sensor Networks

Black and odorous water bodies represent a topic of significant interest in the field of water pollution prevention and control. Remote sensing technology is increasingly exploited for the monitoring of black and odorous water bodies because of its high efficiency and large-scale monitoring potential. In the present study, the Sentinel-2A imagery data were combined with data obtained by measuring spectral properties of black and odorous water bodies to produce a classification and regression tree (CART) model-based improved remote sensing recognition method for such water bodies. This method transforms the traditional single-feature empirical threshold segmentation algorithm to a multi-feature fuzzy decision-tree classification algorithm. The results reveal overall accuracy values of 84.78%, 92.85%, and 72.23% for the CART decision-tree algorithm, the confidence zone classification, and the fuzzy zone node classification, respectively. The method proposed in the present study enables the highly precise extraction of features representing black and odorous water bodies from satellite imagery. The characterization of confidence and fuzzy zones minimizes the need for field inspections, and it enhances the efficiency of diverse applications including engineering.

Cyber resilience in supply chain system security using machine learning for threat predictions

PurposeCyber resilience in cyber supply chain (CSC) systems security has become inevitable as attacks, risks and vulnerabilities increase in real-time critical infrastructure systems with little time for system failures. Cyber resilience approaches ensure the ability of a supply chain system to prepare, absorb, recover and adapt to adverse effects in the complex CPS environment. However, threats within the CSC context can pose a severe disruption to the overall business continuity. The paper aims to use machine learning (ML) techniques to predict threats on cyber supply chain systems, improve cyber resilience that focuses on critical assets and reduce the attack surface.Design/methodology/approachThe approach follows two main cyber resilience design principles that focus on common critical assets and reduce the attack surface for this purpose. ML techniques are applied to various classification algorithms to learn a dataset for performance accuracies and threats predictions based on the CSC resilience design principles. The critical assets include Cyber Digital, Cyber Physical and physical elements. We consider Logistic Regression, Decision Tree, Naïve Bayes and Random Forest classification algorithms in a Majority Voting to predicate the results. Finally, we mapped the threats with known attacks for inferences to improve resilience on the critical assets.FindingsThe paper contributes to CSC system resilience based on the understanding and prediction of the threats. The result shows a 70% performance accuracy for the threat prediction with cyber resilience design principles that focus on critical assets and controls and reduce the threat.Research limitations/implicationsTherefore, there is a need to understand and predicate the threat so that appropriate control actions can ensure system resilience. However, due to the invincibility and dynamic nature of cyber attacks, there are limited controls and attributions. This poses serious implications for cyber supply chain systems and its cascading impacts.Practical implicationsML techniques are used on a dataset to analyse and predict the threats based on the CSC resilience design principles.Social implicationsThere are no social implications rather it has serious implications for organizations and third-party vendors.Originality/valueThe originality of the paper lies in the fact that cyber resilience design principles that focus on common critical assets are used including Cyber Digital, Cyber Physical and physical elements to determine the attack surface. ML techniques are applied to various classification algorithms to learn a dataset for performance accuracies and threats predictions based on the CSC resilience design principles to reduce the attack surface for this purpose.

Task-based online entrepreneurship teaching effect evaluation algorithm based on decision tree classification algorithm

Task-based online entrepreneurship teaching effect evaluation algorithm based on decision tree classification algorithm

A Machine Learning Method for Classification of Cervical Cancer

Cervical cancer is one of the leading causes of premature mortality among women worldwide and more than 85% of these deaths are in developing countries. There are several risk factors associated with cervical cancer. In this paper, we developed a predictive model for predicting the outcome of patients with cervical cancer, given risk patterns from individual medical records and preliminary screening. This work presents a decision tree (DT) classification algorithm to analyze the risk factors of cervical cancer. Recursive feature elimination (RFE) and least absolute shrinkage and selection operator (LASSO) feature selection techniques were fully explored to determine the most important attributes for cervical cancer prediction. The dataset employed here contains missing values and is highly imbalanced. Therefore, a combination of under and oversampling techniques called SMOTETomek was employed. A comparative analysis of the proposed model has been performed to show the effectiveness of feature selection and class imbalance based on the classifier’s accuracy, sensitivity, and specificity. The DT with the selected features from RFE and SMOTETomek has better results with an accuracy of 98.72% and sensitivity of 100%. DT classifier is shown to have better performance in handling classification problems when the features are reduced, and the problem of high class imbalance is addressed.

Algorithm Design for a Cytokine Release Assay of Antigen-Specific In Vitro Stimuli of Circulating Leukocytes to Classify Leprosy Patients and Household Contacts.

BackgroundImmunological biomarkers have often been used as a complementary approach to support clinical diagnosis in several infectious diseases. The lack of commercially available laboratory tests for conclusive early diagnosis of leprosy has motivated the search for novel methods for accurate diagnosis. In the present study, we describe an integrated analysis of a cytokine release assay using a machine learning approach to create a decision tree algorithm. This algorithm was used to classify leprosy clinical forms and monitor household contacts.MethodsA model of Mycobacterium leprae antigen-specific in vitro assay with subsequent cytokine measurements by enzyme-linked immunosorbent assay was employed to measure the levels of tumor necrosis factor (TNF), interferon-γ, interleukin 4, and interleukin 10 (IL-10) in culture supernatants of peripheral blood mononuclear cells from patients with leprosy, healthy controls, and household contacts. Receiver operating characteristic curve analysis was carried out to define each cytokine’s global accuracy and performance indices to identify clinical subgroups.ResultsData demonstrated that TNF (control culture [CC]: AUC = 0.72; antigen-stimulated culture [Ml]: AUC = 0.80) and IL-10 (CC: AUC = 0.77; Ml: AUC = 0.71) were the most accurate biomarkers to classify subgroups of household contacts and patients with leprosy, respectively. Decision tree classifier algorithms for TNF analysis categorized subgroups of household contacts according to the operational classification with moderate accuracy (CC: 79% [48/61]; Ml: 84% [51/61]). Additionally, IL-10 analysis categorized leprosy patients’ subgroups with moderate accuracy (CC: 73% [22/30] and Ml: 70% [21/30]).ConclusionsTogether, our findings demonstrated that a cytokine release assay is a promising method to complement clinical diagnosis, ultimately contributing to effective control of the disease.

Identifying the roots of inequality of opportunity in South Korea by application of algorithmic approaches

This study identifies the roots of inequality of opportunity in South Korea by applying algorithmic approaches to survey data. In contrast to extant studies, we identify the roots of inequality of opportunity by estimating the importance of variables, interpreting the estimated results, and analyzing the importance of individual variables, instead of measuring inequality of opportunity. We apply a decision tree classification algorithm, light gradient boosting machine, and SHapley Additive exPlanations to estimate the importance of the studied variables and interpret the estimated results. According to the estimated results, the region where the individuals grew up, their gender, and their father’s job during their childhood were the main factors contributing to inequality of opportunity. This study proves that the considerable regional disparity and social environment perpetuate gender inequality in South Korean society. It argues that an individual’s socio-economic achievements are strongly influenced by their father’s background, thus, outweighing other family background-related factors. Individuals receive unequal opportunities owing to a combination of region, father’s background, and their own gender, thereby, affecting their socioeconomic achievements. If these factors remain influential from birth to adulthood, removing the conditions that structure them would be one way to achieve equality of opportunity.

Comparison Decision Tree and Logistic Regression Machine Learning Classification Algorithms to determine Covid-19

Many people today are unsure whether they have COVID-19. The frequent fever, dry cough, and sore throat are all signs and symptoms of COVID-19. If a person has signs or symptoms of coronavirus disease 2019 (COVID-19), he/she should see the doctor or go to a clinic as soon as possible. As a result, it's vital to learn and comprehend the fundamental differences. COVID-19 can cause a wide range of symptoms. The experiments were carried out using two Machine Learning Classification Algorithms, namely Decision Tree (DT) and Logistic Regression (LR). Both algorithms were written and analyzed using the Python program in Jupyter Notebook 6.4.5. From the results obtained in the experiments of covid symptoms dataset, on average, the DT model has obtained the best cross-validation average and the testing performance average compared to the LR machine learning models. For cross-validation results, the DT model has achieved an accuracy of 98.0%. For performance testing, the DT model has achieved an accuracy of 98.0%. The LR has obtained the second-best result on the average of cross-validation performance and the testing results. For cross-validation results, the LR model has achieved an accuracy of 96.0%. For performance testing, the LR model has achieved an accuracy of 97.0%. Consequently, the DT for the COVID-19 symptoms dataset is outperforming the LR for cross-validation and testing results.

Nesnelerin İnterneti Cihazlarına Karşı Yapılan Makine Öğrenmesi Saldırıları

As the number of Internet of Things (IoT) devices increases day by day, attacks against these devices are also increasing. In this study, methods of ensuring security in IoT devices and attacks on IoT devices are discussed, and the importance of zero-trust architecture in ensuring IoT security is explained. In addition, the defense rates of padding methods against machine learning used by the attacker are shown and the defense methods used with machine learning techniques are explained. For this purpose, machine learning methods that are effective on attacks, attacks and violations that are achieved by machine learning techniques are specified. In addition, the effectiveness of machine learning techniques in classifying IoT devices in encrypted traffic is examined. The effectiveness of Random Forest and Decision Tree classification algorithms in classifying IoT devices are evaluated. Finally, experiments are carried out for commonly used attack and defense methods. For this purpose, the accuracy rates of the padded and unpadded experiments are compared by analyzing the IoT device traffic. When classifying unpadded data, 84% accuracy rate of IoT devices is achieved, while this accuracy rate has been reduced to 19% with the random padding method that aims to reduce the attacker's rate of accessing correct information.

TWESPA: A MODERN APPROACH TO TWITTER SENTIMENT ANALYSIS

Sentiment analysis has become a stimulating field for both research and industrial domains. The expression sentiment refers to the emotions or thought of the person across some certain issues. Furthermore, it's also considered an immediate application for opinion mining. the large amount of tweets jotted down daily makes Twitter an upscale source of textual data and one among the foremost essential data volumes; therefore, this data has different aims, like business, industrial or social aims consistent with the info requirement and needed processing. Actually, the quantity of knowledge, which is very large, grows rapidly per second and this is often called big data which needs special processing techniques and high computational power so as to perform the specified mining tasks. During this work, we perform a sentiment analysis with the assistance of PySpark framework, an interface for Apache Spark in Python which is taken into account an open source distributed processing platform which utilizes distributed memory abstraction. The goal of using PySpark is that we can run applications parallelly on the distributed cluster (multiple nodes). The effectiveness of our proposed approach is proved against other approaches achieving better classification results when using Naïve Bayes, Logistic Regression and Decision trees classification algorithms. Finally, our solution estimates the performance of Apache Spark concerning its scalability.

Prediction of septic and hypovolemic shock in intensive care unit patients using machine learning.

To create and validate a model for predicting septic or hypovolemic shock from easily obtainable variables collected from patients at admission to an intensive care unit. A predictive modeling study with concurrent cohort data was conducted in a hospital in the interior of northeastern Brazil. Patients aged 18 years or older who were not using vasoactive drugs on the day of admission and were hospitalized from November 2020 to July 2021 were included. The Decision Tree, Random Forest, AdaBoost, Gradient Boosting and XGBoost classification algorithms were tested for use in building the model. The validation method used was k-fold cross validation. The evaluation metrics used were recall, precision and area under the Receiver Operating Characteristic curve. A total of 720 patients were used to create and validate the model. The models showed high predictive capacity with areas under the Receiver Operating Characteristic curve of 0.979; 0.999; 0.980; 0.998 and 1.00 for the Decision Tree, Random Forest, AdaBoost, Gradient Boosting and XGBoost algorithms, respectively. The predictive model created and validated showed a high ability to predict septic and hypovolemic shock from the time of admission of patients to the intensive care unit.

Research on personal credit risk evaluation based on XGBoost

In the current research field of personal credit risk assessment, the existing characteristics and their combination of characteristics have little relationship with the borrower’s default status, and the explanatory ability is not strong. In this paper, feature selection is carried out by Logistc regression, and default discrimination is carried out by XGBoost, and personal credit risk evaluation model is constructed. In addition, 10000 pieces of real credit data of X Bank are demonstrated to test the effectiveness of the model. There are two main tasks in this paper. First, by comparing three feature selection methods, Logistc regression, AIC-Logistic regression and BIC-Logistic regression, the Logistc regression model with the best AUC and accuracy was selected to construct an index system of personal credit risk evaluation. /*Second, by comparing the performance evaluation indexes of XGBoost, decision tree and K-nearest neighbor classification algorithms, XGBoost is selected to build a personal credit risk evaluation model with good performance in AUC, accuracy and Type II error.

Performance Comparison of Machine Learning Algorithms for Albanian News articles

This paper presented a methodology to classify news articles in Albanian languages with machine learning algorithms and NLP. In this proposed work Multinomial Naïve Bayes, Logistic Regression, k-nearest neighbors, Bernoulli, Centroid, SVM, SGD, Perceptron, Passive Aggressive, Decision Tree, and Random Forest machine learning Classification algorithms are trained to classify Albanian news articles and compared in term of accuracy, training time and testing time. We have used 70% of the dataset for training and 30% for tests. The execution time of training and testing algorithms is recorded and presented for different input sizes of data. For experimental we used different numbers of inputs, starting from 8000 articles, and continuing in increasing order by 8000 until the level of 80000 in 8 categories. Experimental results indicate that the Passive Aggressive algorithm shows the best performance in terms of accuracy for Albanian news articles. While the classifier with the lowest accuracy was Random Forest. Logistic Regression, SVM, and Decision Tree have a high delay in training data. SVM also has the longest testing time compared to other classifiers.

Development of a weed detection system using machine learning and neural network algorithms

The detection of weeds at the stages of cultivation is very important for detecting and preventing plant diseases and eliminating significant crop losses, and traditional methods of performing this process require large costs and human resources, in addition to exposing workers to the risk of contamination with harmful chemicals. To solve the above tasks, also in order to save herbicides and pesticides, to obtain environmentally friendly products, a program for detecting agricultural pests using the classical K-Nearest Neighbors, Random Forest and Decision Tree algorithms, as well as YOLOv5 neural network, is proposed. After analyzing the geographical areas of the country, from the images of the collected weeds, a proprietary database with more than 1000 images for each class was formed. A brief review of the researchers' scientific papers describing the methods they developed for identifying, classifying and discriminating weeds based on machine learning algorithms, convolutional neural networks and deep learning algorithms is given. As a result of the research, a weed detection system based on the YOLOv5 architecture was developed and quality estimates of the above algorithms were obtained. According to the results of the assessment, the accuracy of weed detection by the K-Nearest Neighbors, Random Forest and Decision Tree classifiers was 83.3 %, 87.5 %, and 80 %. Due to the fact that the images of weeds of each species differ in resolution and level of illumination, the results of the neural network have corresponding indicators in the intervals of 0.82–0.92 for each class. Quantitative results obtained on real data demonstrate that the proposed approach can provide good results in classifying low-resolution images of weeds.

Potential Factors That Trigger the Suspension of Calcium Carbonate Sediments and Whiting in a Semi-Enclosed Gulf

Whitings, the manifestation of high levels of suspended fine-grained calcium carbonate particles in the water, have been reported and studied worldwide. However, the triggering mechanism of whiting occurrences remains uncertain. The current study attempted to analyze potential factors that might account for whiting occurrences in a semi-enclosed gulf (namely the Arabian/Persian Gulf, hereinafter called the Gulf). First, spatial and temporal variability of whiting events and different potential driving factors (i.e., whiting seasonality, wind-induced mixing, sea surface temperature, and bathymetry) were explored and examined for five years (2015–2020). Second, as a general indicator of whiting occurrences in the Gulf, a whiting index (WI) was developed using time-series analysis and decision tree (DT) classification algorithm. Third, the correlation between the proposed WI and the spatial coverage of various whiting events was examined. Time-series analysis showed that whiting events during the winter season are associated with high winds that lasted for several days. Nevertheless, whiting events were rarely observed despite high wind speed and increased potential for CaCO3 precipitation in summer. This finding suggests that wind-driven forces might be potential sources for mixing water columns, resuspension of CaCO3 particles, and the appearance of whiting in the Gulf. The DT classification algorithm demonstrated that a minimum WI value of 1.1 can explain the initiation of most summer and winter whiting events. Furthermore, a Pearson correlation coefficient of 0.73 was measured between WI and the extent of whiting along the UAE and Qatar coastlines in the Gulf. The proposed WI shows a simple yet effective method for identifying and estimating the extent of whiting in the Gulf.

Prediction of massive bleeding in pancreatic surgery based on preoperative patient characteristics using a decision tree.

Massive intraoperative blood loss (IBL) negatively influence outcomes after surgery for pancreatic ductal adenocarcinoma (PDAC). However, few data or predictive models are available for the identification of patients with a high risk for massive IBL. This study aimed to build a model for massive IBL prediction using a decision tree algorithm, which is one machine learning method. One hundred and seventy-five patients undergoing curative surgery for resectable PDAC at our facility between January 2007 and October 2020 were allocated to training (n = 128) and testing (n = 47) sets. Using the preoperatively available data of the patients (34 variables), we built a decision tree classification algorithm. Of the 175 patients, massive IBL occurred in 88 patients (50.3%). Binary logistic regression analysis indicated that alanine aminotransferase and distal pancreatectomy were significant predictors of massive IBL occurrence with an overall correct prediction rate of 70.3%. Decision tree analysis automatically selected 14 predictive variables. The best predictor was the surgical procedure. Though massive IBL was not common, the outcome of patients with distal pancreatectomy was secondarily split by glutamyl transpeptidase. Among patients who underwent PD (n = 83), diabetes mellitus (DM) was selected as the variable in the second split. Of the 21 patients with DM, massive IBL occurred in 85.7%. Decision tree sensitivity was 98.5% in the training data set and 100% in the testing data set. Our findings suggested that a decision tree can provide a new potential approach to predict massive IBL in surgery for resectable PDAC.

Application of Data Mining Technology in Public Welfare Sports Education in the Era of Artificial Intelligence

Data mining refers to extracting the implicit prediction information from a massive dataset. It has very application prospects. Some data mining tools can develop things. The purpose of this article mainly discusses the public welfare sports education in the artificial intelligence era. The article discusses the research background and significance, development of education data mining, and decision tree technology and enumerates the application of education data mining in real life. The concept of educational data mining is given, and several common typical decision tree algorithms and their connections and differences are described; then, the concepts of multivalue decision tables and decision trees are discussed in detail. This article aims to build a nonprofit physical education system to manage and analyze the attendance data of students’ physical health assessment, so as to improve the enthusiasm of students to exercise, such as BP neural network, decision tree classification algorithm, and cluster analysis, discusses the calculation and analysis process of the relevant body side data of the sports teaching platform, and emphatically discusses and analyzes the application effect of data mining technology in public welfare sports teaching. In addition, this article has built a public welfare physical education system, allowing us to clearly understand various factors that affect students’ exercise and the relationship between various project indicators. Based on these data, educators can adjust technical means. Experimental results can efficiently and conveniently understand the pass rate of students in various sports. The pass rates of students in the six tests, grip strength, and sitting forward bending were 58%, 65%, 78%, 78%, 85%, and 65%, respectively. Using mathematical methods and computer technology, we can dig out valuable education management information from massive education data, so as to provide a reference for improving school enrollment.

Combining Metabolomics and Interpretable Machine Learning to Reveal Plasma Metabolic Profiling and Biological Correlates of Alcohol-Dependent Inpatients: What About Tryptophan Metabolism Regulation?

Alcohol dependence (AD) is a condition of alcohol use disorder in which the drinkers frequently develop emotional symptoms associated with a continuous alcohol intake. AD characterized by metabolic disturbances can be quantitatively analyzed by metabolomics to identify the alterations in metabolic pathways. This study aimed to: i) compare the plasma metabolic profiling between healthy and AD-diagnosed individuals to reveal the altered metabolic profiles in AD, and ii) identify potential biological correlates of alcohol-dependent inpatients based on metabolomics and interpretable machine learning. Plasma samples were obtained from healthy (n = 42) and AD-diagnosed individuals (n = 43). The plasma metabolic differences between them were investigated using liquid chromatography-tandem mass spectrometry (AB SCIEX® QTRAP 4500 system) in different electrospray ionization modes with scheduled multiple reaction monitoring scans. In total, 59 and 52 compounds were semi-quantitatively measured in positive and negative ionization modes, respectively. In addition, 39 metabolites were identified as important variables to contribute to the classifications using an orthogonal partial least squares-discriminant analysis (OPLS-DA) (VIP > 1) and also significantly different between healthy and AD-diagnosed individuals using univariate analysis (p-value < 0.05 and false discovery rate < 0.05). Among the identified metabolites, indole-3-carboxylic acid, quinolinic acid, hydroxy-tryptophan, and serotonin were involved in the tryptophan metabolism along the indole, kynurenine, and serotonin pathways. Metabolic pathway analysis revealed significant changes or imbalances in alanine, aspartate, glutamate metabolism, which was possibly the main altered pathway related to AD. Tryptophan metabolism interactively influenced other metabolic pathways, such as nicotinate and nicotinamide metabolism. Furthermore, among the OPLS-DA-identified metabolites, normetanephrine and ascorbic acid were demonstrated as suitable biological correlates of AD inpatients from our model using an interpretable, supervised decision tree classifier algorithm. These findings indicate that the discriminatory metabolic profiles between healthy and AD-diagnosed individuals may benefit researchers in illustrating the underlying molecular mechanisms of AD. This study also highlights the approach of combining metabolomics and interpretable machine learning as a valuable tool to uncover potential biological correlates. Future studies should focus on the global analysis of the possible roles of these differential metabolites and disordered metabolic pathways in the pathophysiology of AD.

An advanced intelligence system in customer online shopping behavior and satisfaction analysis

Online shopping or internet shopping is increasing day by day. With the advancement of modern technology, the online market is growing in a vast way. People nowadays prefer online shopping because it saves time, energy, and money. Because of the blessing of the internet that online shopping has made its debut which also affects the common citizens for online shopping. So, for the emerging growth of the online market, it is necessary to find out the behavior of online shopping and customer satisfaction. Safety, trust, product quality plays an important role in customer satisfaction. In this study, we examined customer online shopping satisfaction its impact. The quality of the product, the price of the product compared to the local market, the policy of return, timely delivery of the product are also essential elements of online shopping. By analyzing all these factors we have tried to find the customer behavior and satisfaction with online shopping. In our study, we used a machine learning method to search for the result. We used 40 thousand data to find out the accuracy of our work and to analyze customer shopping satisfaction. We use Naïve Bayes, Apiorir, Decision Tree, and Random Forest classification algorithms for this analysis. We have got our best result by using Apiorir algorithm (88% accuracy) and Naïve Bayes algorithm (87% accuracy). We have also focused on customer behavior and interest in online shopping. Our study can help to develop the business intelligence and satisfaction enhancement about E-commerce.

Detection of four alcohol homologue gases by ZnO gas sensor in dynamic interval temperature modulation mode

Metal oxide semiconductor (MOS) gas sensors have poor selectivity, especially in volatile organic compounds (VOCs). Dynamic measurement method of gas sensors can bring a potential to resolve this problem. However, it is also a challenge to distinguish homologue gases with the same functional group. In this paper, four alcohol homologue gases were detected by a ZnO gas sensor. The period of triangular wave is changed to explore the optimal temperature range in dynamic interval temperature modulation mode for detecting ethyl alcohol, n-propyl alcohol, isopropyl alcohol and n-butyl alcohol. The experimental results show that only when the low temperature is lower than 180 °C and the high temperature is higher than 460 °C, the unsaturation phenomenon can be solved, so as to realize the qualitative and quantitative analysis of the four alcohol homologue gases. Under this experimental condition, the concentration gradient measurement of 100–400 ppm is carried out. After the optimization with the decision tree classification algorithm, the recognition accuracy of the four alcohol homologue gases is 97.62%.

Predicting Common Diseases Among Students using Decision Tree (J48) Classification Algorithm

Predictive analysis is very useful in the process of decision making. It discovers useful information by predicting the future outcome. Nevertheless, it is essential to understand an appropriate technique before the predictive analytical model should be developed. This research will compare two predictive methods which are decision tree technique (using J48 algorithm) and rule induction technique (using JRip algorithm). The aim of this research is to build the predictive model for health datasets of students in one of the universities in Selangor. By analyzing medical profiles such as gender, diseases, the symptoms of the diseases, the organs of the diseases and the body systems of diseases, the model can predict the likelihood of disease that may occur in the future. It can offer significant and important insights, such as the patterns and relationships between medical attributes related to the diagnosis datasets. In this study, we are also able to identify the most common illness that may have infected the students in the past five years. This data analysis could be beneficial specifically to the health center to plan, coordinate tasks and make better decisions.