Data visualization and modelling of a smoke detector dataset

• Main Author: Kaniappan, Kathiresan
• Format: Final Year Project / Dissertation / Thesis
• Published: 2024
• Subjects: L Education (General); T Technology (General)
• Online Access: http://eprints.utar.edu.my/6640/1/fyp_CS_2024_KK.pdf; http://eprints.utar.edu.my/6640/

Machine learning (ML) analytics dashboard is powerful tools to monitor, analyze, and communicate the results of your data-driven projects. It can help to track key metrics, visualize trends, identify outliers, and share insights with stakeholders about the dataset. This report aims to continue the progress of the first project report by building on the pre-processing techniques decided on in that report (Flooring and Capping) to prepare a smoke detector dataset for machine learning modelling. After testing out 3 varying models and analysing the output and results, it is decided that the Multilayer Perceptron Model (MLP) has the best performance out of all the models (approx. 92%+ accuracy), also when comparing it to the benchmark model. Furthermore, the output of the model and the model itself has been imported to PowerBI. A combination of Python scripts and PowerBI visualizations has been used to visualize the data in a comprehensible and informative manner showcasing information of the dataset, model performance and key attributes.