2821 research outputs found
Sort by
Optimizing the accuracy of patients length of stay (LOS) in hospitals using Novel Enhanced Gradient Boosting (NEGB) Algorithm in comparison with K-Nearest Neighbors (KNN)
Full text linkThe comparative study is performed on improved
Novel Enhanced Gradient Boosting Algorithm (NEGB) and KNearest Neighbors (KNN) Algorithm, which determines the
patient’s duration of staying in hospitals. The Novel Enhanced
Gradient Boosting by taking the highest priority of maximized
gain is proposed. The proposed algorithm implements Novel
Enhanced Gradient Boosting framework with maximized gain
probability. The Novel Enhanced Gradient Boosting algorithm
and K-Nearest Neighbors Algorithm used in this research
article. The concrete evidence for the performance evaluation,
the sample size prejudge mental tools were used under the
conditions of gpower 0.8, alpha 0.05 and confidence interval
95% during the estimation. The NEGB has a better accuracy
rate (89.90%) compared to KNN has the accuracy rate
(74.25%). The significance of 0.241 (p>0.05) value implies that
experiment driven in this research was insignificant. The NEGB
Algorithm is good in recalculating the stay duration of patients
and also improves accuracy more than the K-Nearest Neighbors
algorithm
DDoS attack prediction using decision tree and random forest algorithms
No full textThe most common network attacks are Denial of Service (DoS) and Distributed Denial
of Service (DDoS) attacks which causes packet loss by delaying the exchange of information,
thereby altering the data packets sent through networks which affect the integrity and reliability
of the data. Over time, various machine learning models have been identified and presented by
researchers to predict and prevent DoS and DDoS attacks. Many researchers have proposed and
used different machine learning techniques to predict DoS and DDoS attacks, however, there is
still a need for improvement in the accuracy of prediction and more evaluation of these algorithms
and a need for more algorithms to be explored. Hence, this paper improves on existing works by
re-evaluating and comparing the accuracy between Decision Tree and Random Forest Algorithms
in predicting DDoS attacks. The results of the paper show that Random Forest (RF) Regression
model is the best-fit model for the cleaned DDoS SDN dataset used because it is more accurate
as it has a lesser mean squared error of 0.21091041940417007 for the test data compared to the
mean squared error value of Decision Tree Regression (DTR) Model. Hence, the paper concludes
that the RF model is the best-fit model to be used in predicting DDoS attacks. However, the paper
proposes that more machine learning algorithms should be explored, implemented, and re-evaluated in detecting DDoS attacks
Technological innovation for digital supply chains within small and medium size manufacturing enterprises
Full text linkThe rapidly growing world of digitalisation opens the doorway for the new era of
automation that plays a crucial role within the industry. Furthermore, technological
innovations that are emerging every day are disrupting traditional business processes
especially within small and medium size manufacturing enterprises (SMEs). The current
industrial revolution pioneer for profit maximisation with cost reduction shows a
significant refinement in improving sustainability that drives forward digitalisation.
Evidence shows that industries have identified digitalisation as a priority in the upcoming
years as the global supply chain is equipping itself with the digital world in the current
industrial revolution. Economic growth is dependent on SMEs around the world where
small and medium size Manufacturing Enterprises (SMMEs) play a vital role in the
current competitive world while they are not able to manage their supply chains
effectively and efficiently due to a lack of optimisation of digitalisation. They identify
that technological innovation is evident for transforming themselves with digital supply
chain, while global market leading organisations are positioning themselves with the
world of digitalisation to their end consumers in their supply chain utilising technological
innovation virtually driving towards a new era of a digital ecosystem.
This research aims to investigate the impact of technological innovation to foster and
promote digital supply chain within SMMEs. Due to the exploratory nature, this study
adopted a case study approach where the data is collected using a semi-structured
interview across 4 cases from three various countries. The findings indicate a lack of
framework for the digitalisation of supply chains within SMMEs, in addition to a lack of
technological innovation and financial constraints that served as limiting factors for
digitalisation of supply chain within organisations. Further, a framework has been
developed consisting of five elements that have been identified from empirical data as
being critical for Digital Supply Chain (DSC) transformation. The theoretical
contributions of this research are the identification of problems faced, limitations of
technological innovation, and an improved understanding of how digital supply chain
transformation can be initiated and achieved in the context of SMMEs. The practical
contribution of this study is imbedded in the developed framework in the form of
recommended strategies for SMMEs for digitalisation of supply chain
Evaluating the performance of a hybrid model for classification of bicycle crash severity and identification of associated risk factors
No full textThis study conducted an exploratory analysis of bicycle crash data from Great Britain with the aim of identifying the key variables that influence the classification of such incidents. It also analysed data on a range of factors that may contribute to bicycle crashes, including the age of the cyclist, lighting conditions, weather conditions, road types, road conditions, and speed limits. Results indicated that these variables are among the most significant predictors of bicycle crashes, with road conditions, time of day, and lighting conditions being particularly vital factors. In addition, the study sought to compare the efficacy of different machine learning and deep learning models in predicting the severity of such incidents. Results indicated that these models demonstrated poor performance in predicting the severity of bicycle crashes. As a result, a hybrid model that combines the K-Nearest Neighbor and eXtreme Gradient Boosting algorithms was developed to improve accuracy. The hybrid model outperformed all other models, achieving an accuracy rate of 83.56%. The study, additionally, has put forward several recommendations, including the mandatory use of reflective clothing and the installation of Intelligent Transportation Systems (ITS) to enhance the safety of cyclists
Peter Bullimore: A living tribute
Full text linkThe aim of this paper is to provide a living tribute of the mental health activist and international trainer Peter Bullimore.
Peter provided a list of people who he wanted to provide tributes. Jerome approached all these people. All agreed.
Several people from around the world attest to the influence that Peter's teaching and personality has had on their clinical practice and on their lives.
The disappearance of Open Mind has left a shortage of journals which welcome the user perspective. Mental Health and Social Inclusion has always championed the voice of people with lived experience. These are selected tributes to one man’s work in the field of mental health.
These accounts provide insights into the work of a remarkable individual.
Students of the mental health professions are mainly exposed to work produced by their peers. The history of mental health is filled with the stories of professionals, not the people who have used services.
Historically accounts of psychiatry are written by mental health professionals. Service user or lived experience accounts are often written from the perspective of the person’s story of illness and recovery. There are comparatively few which celebrate the additional achievements of specific individuals with lived experience
Enhanced preprocessing approach using ensemble machine learning algorithms for detecting liver disease
Full text linkThere has been a sharp increase in liver disease globally, and many people are dying without even knowing that they have it. As a result of its limited symptoms, it is extremely difficult to detect liver disease until the very last stage. In the event of early detection, patients can begin treatment earlier, thereby saving their lives. It has become increasingly popular to use ensemble learning algorithms since they perform better than traditional machine learning algorithms. In this context, this paper proposes a novel architecture based on ensemble learning and enhanced preprocessing to predict liver disease using the Indian Liver Patient Dataset (ILPD). Six ensemble learning algorithms are applied to the ILPD, and their results are compared to those obtained with existing studies. The proposed model uses several data preprocessing methods, such as data balancing, feature scaling, and feature selection, to improve the accuracy with appropriate imputations. Multivariate imputation is applied to fill in missing values. On skewed columns, log1p transformation was applied, along with standardization, min–max scaling, maximum absolute scaling, and robust scaling techniques. The selection of features is carried out based on several methods including univariate selection, feature importance, and correlation matrix. These enhanced preprocessed data are trained on Gradient boosting, XGBoost, Bagging, Random Forest, Extra Tree, and Stacking ensemble learning algorithms. The results of the six models were compared with each other, as well as with the models used in other research works. The proposed model using extra tree classifier and random forest, outperformed the other methods with the highest testing accuracy of 91.82% and 86.06%, respectively, portraying our method as a real-world solution for detecting liver disease
Accuracy of water quality prediction using random forest regression and Artificial neural network
No full textWater is life, human and all activities of living things needs water for its
survival. There are growing concerns about the quality of water available for human
consumption and other daily human activities. The pollution of water are traceable to
economic activities, industrialization, urbanization, agricultural activities and other
human activities. These activities also affect human life and water quality. This
concern has necessitated the studies on prediction of water quality and an efficient
water monitoring system. The most common method of ensuring good quality of
water is by testing them in the laboratory, this current method wastes time, not
economical, wastes human power. Many studies on the use of Artificial neural
network and machine learning models to predict the quality of water has been on the
increase, this study creates an Artificial neural network and random forest regression
machine learning model using parameters such as ph., Hardness, Conductivity, Solids,
Conductivity, Turbidity, Portability and does a comparative analysis of the best fit
model and it was determined that the random forest model had the best fit model with
a lower mean squared error of 2.325103283606211 when compared to the Artificial
neural network with a mean squared error of 3.3122911 by determining the model
with the least mean squared error. This study postulates that random forest method is
the best fit model for prediction of water quality when compared to the Artificial
neural network
Diagnosis of diabetes type using Random Forest algorithm and SVM for improving accuracy
No full textTo improve the accuracy in detection of
diabetes type diagnosis using Random Forest Algorithm
and Support Vector Method. A dataset on diabetes was
collected on kaggle.com. Support vector method analysis
was performed on the data using kernel as linear, poly
and rbf with random state = 42 and test size = 0.2. The
Random Forest Algorithm with n_estimators = 100 has
same accuracy with support vector method with kernel
as rbf. When it comes to type diagnosis for patient having
diabetes, the Random Forest Algorithm and Support
Vector Method with kernel rbf can be used unlike other
kernel of Support Vector Method
Proceedings of ICACTCE'23 - The International conference on advances in communication technology and computer engineering : New artificial intelligence and the internet of things based perspective and solutions
No full textThe third International Conference on Advances in Communication Technology and Computer Engineering.
Today, communication technology and computer engineering are intertwined, with advances in one field driving advances in the other, leading to the development of outstanding technologies. This book delves into the latest trends and breakthroughs in the areas of communication, Internet of things, cloud computing, big data, artificial intelligence, and machine learning.
This book discusses challenges and opportunities that arise with the integration of communication technology and computer engineering. In addition, the book examines the ethical and social implications, including issues related to privacy, security, and digital divide and law. We have explored the future direction of these fields and the potential for further breakthroughs and innovations.
The book is intended for a broad audience of undergraduate and graduate students, practicing engineers, and readers without a technical background who have an interest in learning about communication technology and computer engineering
Flame Retardancy Index ( FRI ) for Polymer Materials Ranking
Full text linkIn 2019, we introduced Flame Retardancy Index (FRI) as a universal dimensionless index for the classification of flame-retardant polymer materials (Polymers, 2019, 11(3), 407). FRI simply takes the peak of Heat Release Rate (pHRR), Total Heat Release (THR), and Time-To-Ignition (ti) from cone calorimetry data and quantifies the flame retardancy performance of polymer composites with respect to the blank polymer (the reference sample) on a logarithmic scale, as of Poor (FRI ˂ 100), Good (100 ≤ FRI ˂ 101), or Excellent (FRI ≥ 101). Although initially applied to categorize thermoplastic composites, the versatility of FRI was later verified upon analyzing several sets of data collected from investigations/reports on thermoset composites. Over four years from the time FRI was introduced, we have adequate proof of FRI reliability for polymer materials ranking in terms of flame retardancy performance. Since the mission of FRI was to roughly classify flame-retardant polymer materials, its simplicity of usage and fast performance quantification were highly valued. Herein, we answered the question “does inclusion of additional cone calorimetry parameters, e.g., the time to pHRR (tp), affect the predictability of FRI?”. In this regard, we defined new variants to evaluate classification capability and variation interval of FRI. We also defined the Flammability Index (FI) based on Pyrolysis Combustion Flow Calorimetry (PCFC) data to invite specialists for analysis of the relationship between the FRI and FI, which may deepen our understanding of the flame retardancy mechanisms of the condensed and gas phases