Implementation of predictive analytics for future risk prediction of Malaysian against COVID-19 infection / Muhammad Haziq Jakaria

On the 31st December of 2019, the first cluster of. Later, the unknown disease was identified as Covid-19 which was a highly infectious virus that can cause chronic respiratory disease. After that, this virus spread throughout the world and was later declared a pandemic by the WHO. Malaysia was no exception and was also affected. Therefore, the MoH come out with an initiative to create a prediction model using the SEIR model to predict cases. However, it does not emphasize the prediction based on events such as the MCO implementation. Due to this, the impact of MCO relaxation has caused a drastic increase of covid cases up to 40,000 cases a day. This big figure also comes with a severe consequence to the saf ety of the public. Therefore, this work presented a forecast that can help to identify a unique pattern during the MCO period based on data from 18 March until 9 June of 2020. This project also employs the CRISPDM methodology until the outcomes can be made into a dashboard. Variables such as date, infection, recovery, and fatality numbers are crucial to achieve better accuracy of ARIMA forecasting. With this research, the EDA on covid trends can be plotted and the forecast is also heavily affected by the component of time series. The result also revealed that ARIMA (1, 2, 8), (1, 2, 3) and (0, 3, 3) showed appropriate results. In short, the ARIMA model is a good model to forecast time series-related data with a proper parameter adjustment.