Exploring earnings data for the UK [Open Data]

jamiet — Mon, 11 Feb 2013 13:34:02 GMT

I have a burgeoning interest in the world of Open Data which wikipedia describes as:

Open data is the idea that certain data should be freely available to everyone to use and republish as they wish, without restrictions from copyright, patents or other mechanisms of control. The goals of the open data movement are similar to those of other "Open" movements such as open source, open hardware, open content, and open access. The philosophy behind open data has been long established, but the term "open data" itself is recent, gaining popularity with the rise of the Internet and World Wide Web and, especially, with the launch of open-data government initiatives such as Data.gov.
http://en.wikipedia.org/wiki/Open_data

To that end I follow @LondonDatastore on Twitter as they are actively publishing Open Data pertaining to the city in which I live, London. Four days ago they announced that a new dataset had been released that provided earnings information, not just for London but for the whole country and going back many years too. The provided link, http://data.london.gov.uk/datastore/package/earnings-workplace-borough, brings up a page from where one can download an Excel workbook containing some data. Unfortunately the data in that workbook is not, in my opinion, provided in such a manner that makes it easily explorable (which I have complained about before on this blog); the data is spread over multiple worksheets:

Worse, the data is already aggregated and pivoted:

In other words the workbook does not contain the raw row-level data from which this pivoted data is produced. Thankfully a link (https://www.nomisweb.co.uk/query/construct/summary.asp?mode=construct&version=0&dataset=99) is provided from where the raw data can be downloaded. I found the interface there to be slightly clunky but that’s a minor quibble – that raw data is available should be considered a major boon. I downloaded earnings data for:

Local Authority
Gender
Full Time or Part Time
Year (1999-2012)

and have made that raw data available in a publicly accessible Excel workbook. You can view that workbook online (only a browser required) at AnnualPay_by_YearGenderLocalAuthorityPartorfulltime.xlsx (from there you can also download for your own analysis).

The workbook also provides summaries over the raw data by way of pivot tables and charts. Arguably its clear to see, for example, that a gender imbalance exists although perhaps the gap may be lessening somewhat

Its interesting to note that the average salary for Males dropped off in 2009/2010. Perhaps the economic events of 2008 are the cause for that, checking out data for only City of London (the traditional financial hub of the UK) suggests that may well be the case:

There will be many more nuggets of information available in this data, all we need to do is set the data free so that people can find it for themselves. That is the aim of this blog post so hit the link: AnnualPay_by_YearGenderLocalAuthorityPartorfulltime.xlsx and see what other nuggets you can find! If dataviz is an area that interests you then this is a cracking dataset to explore!

One other point I want to make is that the raw data is provided as a mean average which means that each aggregated figure is a mean average of some mean averages. This isn’t good as it distorts the data as I demonstrate with this simple example:

We have an average salary for both male and female (5000 & 9000 respectively) and the average of those two averages is 7000. However if we take the total “total salary” / total “tally of people asked” (127000 / 23) then the average is quite different – 5521.73.

That’s a large discrepancy even for only two rows of data and it highlights the problem of providing averages rather than the figures that created those averages. To me this is a data quality issue – the raw source data does not provide the requisite level of detail to enable accurate analysis. Quality of data is paramount.

@Jamiet

HTML Maps for DataViz

jamiet — Wed, 17 Oct 2012 15:16:07 GMT

I don’t talk about data visualisation (#dataviz) much on this blog because, well, because its not something I can claim to be particularly knowledgeable about – that doesn’t stop me from having an opinion about it though.

I just stumbled upon an article that compares the media ecosystems of four various technology companies entitled Mapping The Entertainment Ecosystems of Apple, Microsoft, Google & Amazon and apart from being a well-researched and well-written article (not something that the tech press excels in, in my opinion) I was struck by how well the author uses maps to tell a story to the reader. Take the map in this screenshot:

Clicking on one of the four icons at the bottom of the map dynamically changes the shaded areas of the map to indicate which countries that company offers their services. Its aesthetically pleasing but moreover it instantly coveys useful information to me. What I love about it most of all though is that its all pure HTML – I don’t need any poxy Silverlight or Flash plugin to view the maps and interact with them – all I need is an up to date web browser (I use Chrome v22 although I tested using IE9 and it worked fine there too).

This is how I believe data visualisation should be - conveying useful info in a friction-free way. Maps are a great way of achieving that – I just wish more people agreed with me about calendars as a mechanism for doing the same!

@Jamiet

Search results matching tag 'Data Visualisation'

Exploring earnings data for the UK [Open Data]

HTML Maps for DataViz