Comparative Analysis of Machine learning Model for Diabetes Prediction

Diabetes is arguably the worst disease on the planet. It's not simply a disease; it also increases the chance of developing many other illnesses, such as renal problems, heart attacks, and visual impairments. The most difficult thing a physician can do, whether a patient has diabetes or not, is determine how likely a patient is to get the illness in the early stages. Blood glucose, BMI, age, gender, and family history are some of the interrelated elements that contribute to these problems. A range of Machine-Learning (ML) algorithms have been employed to detect and diagnose the illness in order to stop further health problems. Comparing the machine learning models for the diabetes dataset is the main focus of this work. The PIMA dataset used in this study was acquired from the UCI Repository. There are 9 features in the dataset (768 entries): glucose, pregnancies, skin thickness, insulin, BMI, diabetes pedigree function, and outcome. Random Forest, ANN (artificial neural network), Decision Tree, KNN, and Naïve Bayes are a few of the machine learning algorithms that are implemented. Recall, f1 score, accuracy, and precision are the performance metrics used. With an accuracy of over 79% in comparison to the other predictor, the data indicates that the kNN is the most accurate.