Analysis of COVID-19 data for Victoria, Australia - 27 August 2020

Today's Highlights for Victoria data:

• Considerable community transmission rate.

• The number of applied tests is around 14K.

• The number of new cases is slowly decreasing.

• Moderate level of error in the infection rate estimates under given testing. 

• The number of positive cases relative to the number of applied tests is stable.

• The number of people in the hospital is decreasing.

This page aims to give a picture of the latest situation about COVID-19 pandemic in Victoria, Australia. Hopefully, the analyses and inference here help to deepen the understanding of the trends in COVID-19 data.

All the data is gathered from media releases of the Victoria State Government about Coronavirus disease (https://www.dhhs.vic.gov.au/coronavirus) by myself. For each figure, relevant explanations on the dataset and calculations are given. All the contents and figures will be updated on a daily basis as the new data released by the Victoria Government.

Figure 1 shows the number of daily confirmed cases in VIC along with an empirical exponential curve. Daily confirmed cases information is calculated based on the total number of cases announced daily. The trend in the daily number of confirmed cases followed an exponential upward trend for a while, then started to divert from the empirical exponential curve. Then, we got a downward movement with similar numbers of tests. Now, the movement in the daily number of new cases follows a moving average rather than an exponential trend. We have a decreasing trend in the number of new cases currently. To get a picture relative to the number of applied tests please refer to Figure 6. 

Figure 2 has the total number of cases in VIC in the second vertical axis in addition to the daily number of confirmed cases. The total number of confirmed cases has followed its empirical exponential increase for a while. Now, it started to divert from exponential behaviour. 

Figure 3 exhibits the number of active cases which is calculated based on the total number of cases, the number of recovered and the death toll in VIC (active cases = total number of cases - the number of recovered - the number of deaths). Similar to the total number of cases series, the number of active cases did get back to the empirical exponential-like curve. For a while, the number of active cases in the community had been around the mean level of 800. Recently, it significantly decreased.

The growth rate is an important indicator of the progress of the disease defined as the rate of daily growth in the total number of cases. The daily growth rate and rolling averages of the growth rate for the time windows of 2 and 3 are shown in Figure 4. We want all the curves to point down in general. The growth rate has been fluctuating around a mean level greater than 1.

Figure 5 shows the ratio of the number of daily confirmed cases to that of recovered cases along with a linear fit. This ratio turned to an upward trend from an almost stationary behaviour and was fluctuating around the slightly increasing trend. But on the 28th of March, 12th and 14th of July, a few recoveries were reported. These observations create outlier/unusual observations in Figure 5.  This was a very good example of how outliers pull the fitted linear trend toward themselves as seen in Figure 5. We see this series dropped under the mean level and there was a negative trend due to the decreasing infection rate. 

Figure 6 shows the proportion of positive test results within the tested group. Since the number of tests applied was announced after rounding, the results shown in Figures 6, 7, 8, and 10 are somewhat biased. Until the beginning of April, the series in Figure 6 had an increasing trend with an increasing variance which is an indicator of having a random walk behaviour in general. But then the variation dropped. I added error bars to show the impact of the daily number of test on the estimates. Error bars reflect the amount of error calculated based on Cochran (1977, Sampling Techniques, Wiley, p. 75-76). When the number of tests is less than 3000 level, we see an increased variation in the series (and the error bars as well). Figures relative to the number of applied tests also confirm the decreased growth in the cases for Victoria. In relation to the decreasing number of new cases, the daily number of applied tests also decreased. But then the testing criteria have expanded by Victorian Health authorities to apply more tests to get a better picture of the pandemic in Victoria. The number of applied tests was fluctuating between 3000 and 2000 as a possible implication of the expanded testing criteria. However, it dropped to the level of 1000 tests a day. The decreasing number of infected individuals in the population makes it harder to expand the number of tests too. Then, the Victorian Government expanded the testing criteria once again to test 100K individuals in two weeks at the end of the first wave. In the second wave, the number of tests reached out large numbers. It had been fluctuating between 10K and 15K tests a day and it has jumped up to 30K - 35K tests during the testing blits. Currently, the number of tests is around 15K-20K and the number of new cases relative to the number of tests is slowly decreasing. 

Figure 7 gives the growth rate and rolling averages of the growth rate for the time windows of 2 and 3 relative to the daily number of applied tests. This plot now shows a stronger downward trend. The local upward movement caused by the proportional increase in the number of new cases confirmed with a relatively small number of tests. The curves were fluctuating around a downward trend until we had big jumps, deep dives and other big jumps in the relative growth rate. When the number of tests is considered, the jumps look higher than that of the raw numbers. The relative growth rates were started to settle down with the increased testing effort. But level they settled is greater than 1 implying the spread.

Figure 8 shows the cumulative number of cases relative to the number of applied tests and the fitted exponential curve. The overall trend in this series is getting far from the exponential trend. 

Figure 9 shows the cumulative number of confirmed cases and the daily increase in the number of cases in Victoria that may have been acquired through community transmission. The cumulative number series looks like it follows its empirical exponential curve. However, daily numbers show a large variation in the growth of the total number of cases. The variation in the daily number of applied tests is also another source of the noise. Figure 10 gives a clearer picture.

Figure 10 shows the cumulative series in Figure 9 relative to the daily number of applied tests. When the daily number of applied COVID-19 tests is taken into consideration, the increase in the confirmed cases that can be transmitted through the community started to show an increasing trend.

Figure 11 displays the relative increase in the number of community transmissions to the number of applied tests. Increase in the cumulative number of community transmissions relative to the number of applied tests follows an increasing trend.

The app here gives some insight into the timeframe of the pandemic, the number of unreported cases in the community and the importance of isolation.

Stay healthy.

* All the information on this page is based on publicly released data and under the sole responsibility of the author and does not necessarily represent the opinion of any other person or entity.