Stroke Prediction using Optimization and Exploratory Data Analysis

Abstract

A stroke is a medical condition that causes brain damage by rupturing blood vessels. It can also happen if the brain's blood supply and other nutrients are cut off. It is the leading cause of death and disability worldwide., according to the World Health Organization (WHO). It is a potentially fatal illness that primarily affects adults over the age of 65. Doctors devote a significant amount of time and effort to predicting strokes. As a result., the primary goal of the study is to use various Machine Learning approaches to predict the likelihood of stroke occurring using hyper parameter tuning to achieve greater accuracy and optimize the outcomes. After going through the dataset, we discovered that the standard algorithms we used., such as Support Vector Machine (SVC), Decision Tree Classifier, Random Forest Classifier, XGBoost, and KNeighbors, as well as some feature selection methods, could only predict 80 to 85 percent of the time, so we came up with the idea of optimization in machine learning, where we use the technology or concept of hyper parameter tuning, which helped us to gain a prediction of about 95 percent. With this, we also used an Exploratory Data Analysis (EDA) concept for visualization, which helped us to study the attribute. The above-mentioned prognosis was achieved using Hyper Parameter Tuning, which involves checking and analyzing the parameters of each algorithm in such a way that after setting to some predefined parameters, it produces the expected accuracy. To evaluate the data, we employed the EDA approach, in which we compared many associated health behaviors in different combinations with respect to stroke, and each EDA diagram concluded the relationship of these attributes to the cause of stroke. As a result, this study evaluates the performance of various machine learning algorithms that use Hyper parameters tuning with EDA.