COVID-19 Data Analysis 2

This COVID-19 data analysis 2 will focus on the percentage of pneumonia, influenza or COVID-19 deaths to total deaths. This analysis addresses the issues of analyses of total deaths and number of new cases, which can be deceiving for various data-source reasons.

The following analyses applies a 30,000-foot-level process-output metric tracking methodology. This high-level time-series tracking approach analyzes the output of a process will determine if the response is stable, even considering seasonal variation.

If stability exists, the 30,000-foot-level report-out can provide a prediction statement.  If the prediction statement is undesirable, there needs to be a positive change to the inputs to the process to improve the output’s response level.

COVID-19 Data Analysis: Using 30,000-foot-level Reporting and Analysis Techniques

On April 19, 2020, I provided a 30,000-foot-level COVID-19 analyses for US and Texas. The measurements analyzed were reported COVID-19 deaths and new cases.

However, an analysis of COVID-19 pandemic that examines the number of deaths and new cases has issues for several reasons, which include:

It would seem that a better measurement to determine whether the pandemic is peaking or not would be to analyze the percentage of pneumonia, influenza or COVID-19 deaths to total deaths.  If this percentage increased in recent weeks, one could assume that the increase was caused by COVID-19 issues. Similar, if there were a decrease after a rise, then one could assume that there was a decrease in COVID-19 issues.

However, this measurement approach has seasonality issues. Pneumonia and influenza deaths are historically higher in the winter months. The analysis below will examine this percentage value and address seasonality changes.

Source of this COVID-19 Data Analysis 2

The following two tables were taken from the same data source at different times. When comparing these two data tables, one notes that data are updated weekly and can change dramatically for later weeks in the data set from one week report-out to another.

To illustrate this point, 2020 week 16 for the two charts are very different. In the first table, the week 16 value for total deaths is 28,483, while in the second table, this week 16 value is 63,131. Apparently the reason for this difference is that more data is available over time for a newly reported week.

The data analysis below will use the later table values.

Data from

COVID-19 Data Analysis 2, Week 16 Data


Data from

COVID-19 Data Analysis 2 -- Week 18 Data


30,000-foot-level Analysis Percent of Deaths from Pneumonia, Influenza or COVID-19 Deaths

The chart below was staged on 3/29/2020, since it appears that is when the percentages of pneumonia, influenza or COVID-19 deaths to the total deaths increased.

The 30,000-foot-level chart indicates stability; however, from this chart, it is apparent with this limited amount of data that there was a trend upward and now downward. Because of this trending after the stating of the individuals chart, my conclusion is that this metric for COVID-19 is not stable; hence, not predictable.

This is good news in that this COVID-19 metric indicates that the frequency of COVID-19 issues is getting smaller.


COVID-19 Data Analysis 2 -- percentage to total deaths


However, from the individuals chart on the left, it appears that there may have been a trend in deaths before 3/29/2020, which some have conjectured to be true.

Typically winter months have more respiratory problems. Unfortunately this data set did not go back for a complete year. However, to test for stability, one would not expect that there be a trend in the difference between adjacent weeks; hence, we will now examine this delta difference between weeks metric using a 30,000-foot-level chart.

A 30,000-foot-level chart of this difference between week values is:


COVID-19 Data Analysis 2 -- week-to-week measurements


The individuals chart in this analysis did not indicate any trend in between-week percentage rates before about 3/29/2020.

Like from the other 30,000-foot-level report-out, from this chart we could conclude that COVID-19 issues have peaked and things are getting better. Because of this trending after the stating of the individuals chart, my conclusion is that this metric for COVID-19 is not stable; hence, not predictable.

Let’s hope that “opening things up” does not negatively impact this apparent downward trend.


Contact Us to set up a time if you would like to set up a time to discuss this COVID-19 Data Analysis 2 or 30,000-foot-level reporting

Leave a Reply