# Impressions from Data Sketches

By chance, I spotted a tweet this evening from @owenboswarva pointing to a *#DVLA data release: number of licence holders with penalty points, broken down by postcode district [link] | #FOI #opendata*.

A quick search turned up some DfT driving license open data that includes a couple of postcode district related datasets – one giving a count of number of license holders with a particular number of points in each district, one breaking out the number of full and provisional licence holders by gender in each district. The metadata in the sheets also suggests that the datasets are monthly releases, but that doesn’t seem to be reflected by what’s on the data.gov.uk site.

Whatever…

I haven’t done any data sketches for a bit, so I thought I’d have a quick play with the data to see whether any of the Isle of Wight postcode areas seemed have a noticeably higher percentage rate of points holders than other bits of the island, dividing the number of license holders with points by the total number of license holders in each postcode district…

If you’re not familiar with Isle of Wight postal geography, Wikipedia identifies the postcode districts as follows:

PO30 NEWPORT PO31 COWES PO32 EAST COWES PO33 RYDE PO34 SEAVIEW PO35 BEMBRIDGE PO36 SANDOWN PO36 SHANKLIN PO37 SHANKLIN PO38 VENTNOR PO39 TOTLAND BAY PO40 FRESHWATER PO41 YARMOUTH

So Seaview and Yarmouth, then, are places to watch out for (places that tend to have older, well to do populations…)

I then wondered how the number of license holders might compare based on population estimates. From a quick search, I could only find population estimates for postcode districts based on 2011 census figures, which are five years out of date now.

The main thing that jumped out at me was that for the number of license holders to exceed the population means there must have been a noticeable population increase in the area… The really high percentages perhaps also suggest that those areas have older populations (something we could check from the population demographics from the last census). Secondly, we might note that the proportions rank differently to the first table, though again Yarmouth and Seaview head the leaderboard. This got me thinking that there are perhaps age effects making a difference here. This is something we could start to explore in a little more detail using a couple of the other DfT tables, one that describes the number of licenses issued by gender and age, and another that counts the number of points carrying a particular number of points by age and gender. (These two tables are at a national level, though, rather than broken out by postcode district.)

I guess I should really have coloured the numbers using a choropleth map, or using a new trick I learned earlier this week, displayed the numbers on labels located at the postcode district centroid…

(The map shows Land Registry House Prices Paid data for November, 2015.)

Maybe another day…