Back to Projects

Team Name:

Prophet


Team Members:


Evidence of Work

Real-Time Prediction model based on time series in the post-epidemic era

Project Info

Prophet thumbnail

Team Name


Prophet


Team Members


3 members with unpublished profiles.

Project Description


The first confirmed case in Australia was identified on 25 January 2020. The second wave of infections emerged between May and June 2020.

This year, the outbreak of the Delta variant makes almost half of Australia's population and most major cities lockdown.

As we can see although people progress to post-pandemic, people still face the challenge of virus variation and risk of re-opening.

Thus, we hope to use real-time data to predict the number of confirmed cases, which could give the government a real-time warning.

Based on the prediction, the government can know which place should pay attention to. We also use data analytics to find useful suggestions for epidemic management.

To predict the real-time number of confirmed people. We figure out an efficient way by using Timeseries Model and Confluent.

Firstly, we use past data to train a powerful time-series model. Secondly, to update real-time numbers, we use a confluent platform to connect python and output websites.

We input real-time data to confluent, and this data is transferred to the Timeseries model. Our model will predict a value, which will be transferred to the website, which shows the curve of predicted confirmed people.

After prediction, we still need to know what action should be taken to control the epidemic.
We hope to find some factors that are critical to the control of the epidemic

We compared the curves of the level of government intervention with the number of diagnoses over time, and we found that each major outbreak of infection was preceded by a decline in the urgency of government intervention.

This leads us to conclude that governments must not be complacent about epidemic prevention at any time.

Besides, we built a model to judge the relevance between some factors and confirmed cases. The results show that facial coverings have the highest correlation, and increasing facial coverings can significantly prevent people from being infected.

In addition, increasing the containment health index, stricter restrictions on gathering, and closing of workplaces are all relatively effective in controlling the number of COVID-19 infections.


#real-time #prediction #time-series

Data Story


We use the data form Oxford Covid-19 Government Response Tracker (OxCGRT) for time series model building and testing.
Besides, we use the data from Oxford Covid-19 Government Response Tracker (OxCGRT) and Oxford Covid-19 Government Response Tracker (OxCGRT) for data analytic. From these datasets, we extract useful information


Evidence of Work

Video

Project Image

Team DataSets

Oxford Covid-19 Government Response Tracker (OxCGRT)

Description of Use Use it for model building and data analysis

Data Set

Australian National University CRISPER data within Confluent Cloud

Description of Use Use it for model training and data analysis

Data Set

COVID-19 Data for Australia

Description of Use Use it for model training and data analysis

Data Set

Challenge Entries

Living in a post-pandemic world

How can we use real-time data to assist with the preparation and ongoing management of living with Covid-19 as we progress towards a post-pandemic Australia and New Zealand?

Eligibility: Contestants should use Confluent as part of their solution and the CRISPER dataset as part of their solution.

Go to Challenge | 13 teams have entered this challenge.