An obvious way to present time series data extracted from a single table is to graph all the variables on the same graph. This is efficient spatially and, when the variables are drawn from the same range, can result in a reasonable presentation of the data.
An alternative approach is to pull out all the variables and plot them individually. While this obviously takes up more space, it offers a few advantages.
Firstly, if the data is drawn from very different ranges (and, perhaps, just present in the same table as a matter of convenience) then it prevents one data set from skewing the presentation area away from a range appropriate to the other variables.
Secondly, when the number of variables is high, it allows visibility equally to all of them.
Below is an experimental alternative to the initial embedding approach I've explored in other posts backed by d8taplex data. Here I've simply iterated through all the variables in the data set presenting each one as a sparkline and its label.
This data (permalink) shows Household Consumption of Selected Foods from 1942 onwards in the UK, and has been extracted from www.defra.gov.uk.
This is one of my favourite data sets, showing the long term trends of consumption. For example, we can clearly see the up-tick in skimmed milk, and many disturbing downwards trends (fresh green vegetables anyone?).
Comments