The performance comparison of ensemble machine learning classifierson medical datasets

The machine learning process may be broken down into two distinct phases: learning, in which the classification algorithmic program is trained, and classification, in which the algorithmic program labels new data. Classification, also known as supervised learning, is a technique for processing data that divides the information into predetermined groups and categories. We are examining the J48 Decision Tree, the Artificial Neutral Network (ANN), the Extreme Learning Machine (ELM), and other other notable ensemble machine learning classifiers. Single-hidden-layer feedforward neural networks spawned the new field of ELM, which focuses on regression and classification challenges. In this thesis, activation functions will be used to apply the three basic classifiers—ELM, J48, and ANN—to medical datasets. These databases include information on diseases such as diabetes and hepatitis. The ELM classifier is fed both healthy and harmful features. In addition, the logarithmic and tangent sigmoid activation functions, as well as others, are included into ELM in order to optimize classification performance. Based on information theory, the J48 approach does classification using decision trees. It's a version of Ross Quinlan's older ID3 approach, sometimes referred to as J48 in Weka and named after Java. SVM is sometimes referred to as a statistical classifier due to the fact that C4.5 creates decision trees that are utilized for classification. The C4.5 algorithmic rule (J48) is often used to categorize data in a vast array of fields, including the diagnosis of coronary heart disease by evaluating clinical data, classification of data for e-governance, and many others. In the realm of computing, artificial neural networks (ANNs) were developed in an attempt to emulate the neuronal network that makes up a person's brain. ANNs enable computers to absorb information and make judgments in a way that closely resembles that of humans. The artificial neural network is the product of programmers' desire for computers to act similarly to an interconnected network of brain cells. We will employ classifiers such as ELM, Sigmoid, and Relu for a medical dataset of patients with a patient population of 900 to 1000, leveraging a free online dataset of patients with mild to severe illness difficulties.