On the 9th December 2010 in Birmingham there were:

2 enquiries about a public emergency

9 enquiries relating to odour

4 notifications of death (council tax)

1 request for information on the nearest public toilet

In the fine detail of the data emerges the hints of human stories, forming questions in the gap between the reductive nature of a categorised event and the image it forms in our minds. What odours are being enquired about? What are the public emergencies? We may never know the answers, but the data can provide a snapshot of what is happening( or at least being reported as happening) from the total volume and frequency of enquiries to the fine detail of reported events occurring within wards and postcodes.

Percentage of calls by Birmingham constituency on the 9th December 2010

Over the past month we have been wrestling with a large data set of approximately 5000 entries a day. Trying to tease out meaning from the sheer volume of information sloshing around in the tables has been a challenge. At first sight the individual records can seem meaningful, but attaching meaning to the aggregated numbers can be a bemusing process. A number of issues have come up in developing the dashboard (the one that didn’t come up was the access to the data) Birmingham City Council have been very accommodating from the start, particularly Nick Smith who has continuously provided us with data extracts tailored to our changing needs. The issues have emerged not as an issue of access to data but rather how to understand and interpret the data. I have outlined these issues below:

1. Terminology

As with any data set understanding the terminology is key. Each enquiry received is categorised into location and type. The type refers to a set of council directorates and their sub categories (4 levels). The naming of these directorates can be confusing to the outside eye. Titles like Chief Executives don’t really allude to what is contained within, similarly Development and Culture seems such a broad church that practically any council activity could contained within. With this comes layer upon layer of abbreviation that has to be defined and turned into plain english from “services” being abbreviated to “svs” and CYPD – Children, Young People and Families.

2. System Overview

We have been constructing the dashboard from the data up. So the our view of the fields and table’s is based upon the data that is in them. If there is not an enquiry to a particular field we have no idea whether it exists or not. We only know so much about the framework of the whole CRM (contact relationship management) system. So there is always an element of working in the dark.

3. Shifting Sands

When you are dealing with a live dataset that encompasses the total activity of an organisation as complex and mutable as a city council. The chances  of directorate structures to staying the same is very low. Each new data extract we may see a re-shuffle in a department that changes the naming of a council activity or that activity moving to a  new directorate. For a project that is live and tracking council traffic daily this could present a significant challenge.

4. Time

As the maps and graphs assemble in our views of the data, it is difficult to see specific events occurring as the general spread may be to wide the changes to insignificant in a 24 hour period. But once the dashboard starts collecting over days and months, we may need to look at how best to re-factor the way we present the merging trends. Trying to design for the unknown is an interesting process.

We are close to launching the dashboard. However we are now at the mercy of the City Councils’ private contractors who will be producing the daily feed from the CRM. This has been approved as a work order, but when this comes into play is still a matter for how high a priority we are in the council’s IT rollout.

Matt Watkins