Key Datasets
This page contains overlaid datasets, with and without scaling and time shifting. The purpose of this is to look for trends spanning different datasets, to help find evidence for causality and lag between these datasets.
At present, there is no discussion about these graphs; the reader is invited to draw their own conclusions from these data.
Dataset alignments are modified over time as the data evolves; clicking on the legend items for each graph temporarily removes a dataset from view, which is useful when looking for trends over time.
Best Fit
In this graph, all data sets are normalised to fit on the same y-axis, and time-shifted along the x-axis, to create an optimal fit. Doing this adjusts for the differing delays or lags between datasets.
Below we have three version of the graph, the second having a logarithmic y-axis.
The third graph focuses on data from 1stMay 21 onwards. These data have been scaled to more clearly see the causal relationship between infections and hospitalisations. In this graph hospital admissions have been scaled up 3.7 times, and mortality scaled up 10 times.
(NB: t = date the dataset in question is reported up to; t - x indicates the data has been shifted by x days to the left of it's published date in the fitted graph & t + x indicates the data has been shifted by x days to the right of it's published date in the fitted graph).
Logarithmic
Hospital & Symptom Study
Hospital data, mortality, and COVID symptom study data overlaid without time shifting.
Click legend items to hide and reveal datasets:
Admissions & Occupancy Fitted
Click legend items to hide and reveal datasets:
Mortality Fitted
Alternative Fits
Below we have a series of alternative versions of the 'Best Fit' graph above.
Click legend items to hide and reveal datasets: