Page 73 - FULL REPORT 30012024

researcher gained a deeper understanding of stroke types, risk factors, and the

importance of developing methods to counter stroke.

Based on this chapter, the researcher made decisions regarding the suitable

methods and tools for the project. Random Forest (RF) machine learning was
selected for its proven effectiveness in related works, where the classification

model achieved a high accuracy of 96 percent. Python was chosen as the

programming language for its ease of use and the availability of various
libraries for data analysis and visualization. The Anaconda environment and

Jupyter Notebook were utilized throughout the data analytics process. For
data visualization, Power BI and Chart.js were selected due to their wide

range of options, including graphs, maps, and customizable dashboards, and

their ability to integrate seamlessly with different data sources and web
applications.

MongoDB and MySQL were deemed suitable for storing the project's data,

offering high availability, replication capabilities, and efficient management
of large data volumes. The Flask web framework was chosen for building the

web application, providing a straightforward and flexible approach. The input

parameters requested from users included gender, age, marital status,
occupation type, residence area type, BMI, smoking status, and the presence

of hypertension and heart disease. For system testing, the Technology
Acceptance Model (TAM) was used to understand users' acceptance of the

dashboard visualization and adoption of the machine learning technologies.

Overall, these choices ensured that the project benefited from accurate

machine learning algorithms, efficient data analysis and visualization using
Python and Power BI, seamless data storage with MongoDB and MySQL,

and a user-friendly web application development framework in Flask. Figure

2.20 showed the relevance tree based on the literature review chapter.

56