Bertók, CsanádNagy, DávidSingh, Veer2021-11-152021-11-152021-11-15http://hdl.handle.net/2437/324635This thesis tries to create predictive models to predict stroke. We explore three supervised machine learning algorithms; K-Nearest Neighbors, Support Vector Machine, and Random Forest to create these predictive models. The dataset used contains over 5000 instances of patient records with features like age, BMI, marital status, etc. Before training the models, the data was researched using exploratory data analysis methods and then fine-tuned using feature engineering. Finally, the dataset was encoded to allow it to be read by the algorithms. Each model was further improved with hyperparameter tuning, achieving F1-scores of above 96%.57enMachine LearningClassificationSupervised LearningStrokeDEENK Témalista::Informatika