I used a combination of Matplotlib (a graphics package for Python), Flask (micro webframework for Python that has been used by a company as big as LinkedIn) and Heroku (for being able to put the graphs online - before Heroku axed their free tier) to do a variety of graphs earlier in the COVID-19 pandemic.

Accessibility note: given how many people are affected by colour blindness (e.g. 4.5% of UK population) I should have remembered to note in all the following graphs the green line is the triangles and the blue line is the dots (now noted in captions).

Dual axis charts

There are some good reasons to not do dual axis charts - but I wanted to do some and you can do them in Matplotlib (while Hadley Wickham has arguments against them and does not support them in his popular R charting library, ggplot2).

Cases & admissions

Graph shows a hospitalisations peak of over four thousand per day as a 7 day rolling average (January 2021) and a cases peak of one hundred and seventy five thousand cases per day as a 7 day rolling average (January 2022). The graph shows wildly varying ratios of cases to admissions with cases below twelve thousand in 2020 first wave while admissions were three thousand a day to cases around one hundred and seventy five thousand while admissions were just over two thousands a day in January 2022 with the rise of Omicron. The period where the graph shows the most constant ratio is summer and autumn of 2021 with about 1000 admissions a day for 40000 cases a day. Cases and admissions from start of pandemic data to 2022/10/12 (the green line is the triangles and the blue line is the dotted line).

Note that cases data is heavily dependent on testing - which has varied in the pandemic. There was relative scarcity of testing initially - then there was the development of lateral flow tests and their free supply to the public and then that free supply was terminated. So cases to admissions ratio comparisons only make sense within short time periods - when one is sure the testing policy and supply of tests is comparable.

Graph of cases and admissions with a relatively constant ratio for 2022/04/08 to 2022/10/12 with a peak in early July of two thousand admissions while cases where twenty thousand - when admissions being 5 percent of cases was the more normal ratio during period Cases and admissions for the 180 days leading to 2022/10/12 (the green line is the triangles and the blue line is the dotted line).

Cases - by publication date and rolling 7 day average

A more prosaic use of plotting dual axes shows the rolling 7 day average and the cases by publication date.

The rolling 7 day average is never more than fifty thousand but the raw numbers by publication date are over one hundred thousand on the graph on four occasions in the 2022/04/08 to 2022/10/12 period shown as the figures are published weekly This graph shows rolling 7 day average cases for the UK and the figures by publication date - the broad difference being due to new cases being reported on a weekly basis. (The green line is the triangles and the blue line is the dotted line).
Starting at the beginning of the COVID-19 pandemic in UK in 2020 this shows the cases by publication date and the rolling 7 day average are close - until earlier 2022 when the figures are updated weekly which shows as a wildly oscillating green line - which has previous mirrored the blue dotted line of the rolling average quiet closely. This graph shows rolling 7 day average cases for the UK and the figures by publication date - the transition to weekly reporting shows up clearly as a divergence of the rolling average and publication date figures. (The green line is the triangles and the blue line is the dotted line).

The rise of Omicron

As you’d expect the rise of the Omicron variant shows up in the charts.

For autumn 2021 this graph shows steady case numbers for UK in the thirty to forty thousand cases per day range until shortly after the 9th of December 2021 when cases rise sharply to eighty thousand a day by 20th December 2021 In this graph of cases in 135 days to 2021/12/20 one can see the step change occasioned by the spread of the Omicron variant. (The green line is the triangles and the blue line is the dotted line).
In the three months leading to 20th December 2021 this graph shows - for London - the 7 day rolling average for hospital admissions was around 100 per day and the cases between two and four thousand per day - until the first week of December when a sharp upward trend is seen with figures ending at around nine thousands cases a day (for 13th December a week prior to 20th) and nearly 200 admissions on the 20th - cases a week prior predict admissions a week later generally - hence using prior week's admissions in graph. In this graph of the 3 months leading to 2021/12/20 one can see even in the hospital admissions data and cases for London that Omicron is bringing about a rise in December 2021 that is beyond that of having Delta. (The green line is the triangles and the blue line is the dotted line).

Data retrieved from UK government API (data believed to be under Open Government Licence v3.0).
Note that 7d averages seen above were calculated using pandas .rolling(7).mean() method on the dataframe and unlike the UK government dashboard includes figures from previous 5 days (which may be incomplete) - so accuracy for the latest 7d average was sacrificed in favour of including the latest data.
Note also that graphs were the data as available at the time and there may have been revisions (e.g. issues with lab tests).